CANADA - UNITED STATES
Transboundary PM
Science Assessment
>ji\ lr
The report was undertaken by the Canada-U.S.
Subcommittee on Scientific Co-operation, in support of the
Canada-U.S. Air Quality Agreement


-------
CANADA - UNITED STATES
TRANSBOUNDARY PARTICULATE MATTER
SCIENCE ASSESSMENT
A REPORT BY THE
CAN AD A-U.S. AIR QUALITY COMMITTEE
SUBCOMMITTEE 2: SCIENTIFIC COOPERATION

DECEMBER 2Q04

-------














-------





-------
chapter 2
Figure 2.1- Average F>M2 g
concentrations. The U.$;. data
are from FEM monitors at sites
in the EPA AIRS database for
July 1998 through July 20OOj
Canadian data are from TEOM
and dichotomous samplers
operating from 1995 through
2000. The currently available
data from sites in Mexico
represented less than one year
of sampling and were excluded
from the computation of annual
averages.^ Spot diameter varies
in proportion to concentration.
(Sotira:: R. Husar,:pers. comm.):
Annual PM2.5 Mass Concentration
Stations:
AIRSFRM 1007, 1938-OC
Canada TEOM 43,1995-00
Canada Dichot 24,1995-OD
Annual PM2 5, pg/rn3
* :1> 15 '
K 10- 15
. <10
acid deposition focus on wet deposition of SOj
and NO3 , research in south-central Ontario indi-
cates that approximately 40% of the total deposi-
tion of sulphur and nitrogen occurs in the form of
dry deposition (Sirois et al., 2001). These acidify-
ing pollutants have been shown to damage terres-
trial and aquatic ecosystems and susceptible
materials at levels measured frequently in Canada
and the northeastern United States.
The geographic region most affected by acid
deposition is southeastern Canada and the north-
eastern United States; east of Manitoba and south
of 52 degrees latitude. The relative contribution of
the sources of acid deposition (local versus long-
range) is area-dependent, however, the majority of
acid deposition in southeastern Canada originates
from long-range transport, as does a significant
proportion: of the deposition in the northeastern
United States.
2.4 PM AND ITS PRECURSORS
ARE A SIGNIFICANT
CAUSE OF VISIBILITY
IMPAIRMENT.
Optically, PM interferes with visibility by either
absorbing or scattering visible light. Light scatter-
ing is roughly proportional to the mass concentra-
tion of fine particles, while light absorption is
roughly proportional to the mass concentration of
the light-absorbing species. The impairment of
visibility that results from the absorption or scat-
tering of light reduces the distance to which one
can see and decreases the .apparent contrast and
colour of distant objects, causing a washed out or
hazy appearance.
The light extinction effects of PM vary with
particle size, chemical composition, and humidity.
The particles with the greatest influence on visibil-
ity are fine particles of the same scale as the wave-
lengths of visible light (approximately 0.3 to 1 mm
in diameter). These particles are generally com-
posed of SO= and NO3 salts, OC, or BC.
5

-------
Canada - United States Transboundary PM Science Assessment
2.5 PM AND ITS PRECURSORS
CAN BE TRANSPORTED
LONG DISTANCES.
PM can remain in the atmosphere for days to a few
weeks, depending on the size and rate at which it
is removed from the atmosphere (e.g., by precipi-
tation). Particles in any given area may originate
locally or from sources hundreds to thousands of
kilometers away. Particles can also be formed dur-
ing atmospheric transport from precursor gases
originating from either local or long-range sources.
Both local and regional emissions: underlie
local ambient concentrations in many urban areas,
Regional contributions from sources distant from
eastern North American urban sites (including
upwind urban areas) can account for 50 to 75 per-
cent of the total observed PM2 5 mass concentra-
tion within a specific urban area.
2.6 PM AND ITS PRECURSORS
ARE TRANSPORTED
BETWEEN THE UNITED
STATES AND CANADA.
The NARSTO PM Assessment described two studies
in Canada and the United States that demonstrate
the transboundary transport of PM and its
precursors.
Brook et al. (2002) traced 3-day back-trajecto-
ries of air masses arriving at Simcoe, ON during
the warm season (May-September) of 1998 and
1999 (Figure 2.2). These back-trajectories were
divided into categories based on the concentration
of PM2 5 measured at Simcoe and the directionality
of the contributing air mass. This analysis resulted
in three "source-receptor" categories: 1) "low"
PM2 5 (6 hour averages of 6.8 pg/m3) category,
characterized by north-south airflows, 2) "high"
(22.4 pg/m3) PM2 5 category, characterized by
Figure 2.2- 3-day hack-trajectories arriving at Simcoe. Ontario, for the warm season (May^Septemfeerj, 1998 and 1999.
(The sectors shown represent a) northerly flow over predominantly Canadian source regions and b) southerly flow over
U.Ssource regions. Corresponding median PMa g concentrations are a) Sector 1: 3.8 fig/m3 and
b) Sector 2: 20.3 pg/m3).
6



-------
Chapter
3,
AMBIENT OBSERVATIONS IN BORDER REGIONS
3.1 LEVELS OF AND TRENDS
IN PM _
. O
3.1.1 Integrated Observations between
Canada and the United States.
Levels of PM and PM precursors are monitored
and reported across the United States and Canada.
Monitoring techniques vary between the two coun-
tries,, but wherever possible in this Assessment,
efforts have been made to account for differences
in techniques and combine monitoring results to
provide a more comprehensive view of PM levels in
the border regions. Figure 3.1 illustrates mean
annual PM2 s concentrations at Canadian dichoto-
mous (dichot) and U.S. Federal Reference Method
(FRM) sites. Annual levels of PM2 5 are as high as
18 ug/m3 in the northeastern United States, but
are consistently lower than 12 [ig/m3 in the mid-
continental States. The bi-national map in Figure
3.1 shows few monitoring sites; north of the
Canada-U.S. border due to differences in sampling
frequency between the two countries.
When Canadian hourly TEOM observations-
are included in the database, a more detailed pic-
ture of ambient levels can be achieved. The 98th
percentile values for the years 2000-2002 are
shown in Figure 3.2. The northeastern United
States is again a region of high ambient PM2 5
levels,, with 98th percentile values in excess of
30 yg/m3 at a majority of the sites. Canadian loca-
tions- exhibit generally lower levels of PM2 5,
although concentrations greater than 30 pg/m3
occur in several regions of the country for the years
2000-2002, particularly in the Windsor-Quebec City
corridor.
Time trends of gaseous S02, particle SOj, par-
ticle NHj and total nitrate (HN03 + NOj) concen-
trations were investigated at a number of
rural/remote sites in the eastern United States and
Canada from 1989 to 2002 (Figures 3.3 and 3.4).
Canadian measurements were made by the
Canadian Air and Precipitation Monitoring
Network (CAPMoN), and U.S. measurements by
Canada Dichot
and US FRM
PM2.5 Monitors
2000 -2003 mean
10-<=12
12 -<=15
15 ,<=18
y-
Figure 3.1 - Mean annual con-
centration of PMg 5 at
Canadian dichot and U.0, FRM
monitors in the border region
for the data years 2000-200%.
(Notr- 07([i,f< li.:ir sit^s are years
2000-2002 not all sites include
IliO 'e [ nit fears of d ata 1.
9

-------
Canada - United States Transboundary PM Science Assessment
Canada TEOM
and US FRM
PM2.5 p98 00-02
ilelTEOM monitor
the Clean Air Status and Trends Network
(CASTNet). The two networks use similar filter-
pack sampling technology, but the Canadian meas-
urements are 24-hour average concentrations-
while the U.S. measurements are weekly-average
concentrations. This difference has no significant
impact on the comparability of the trends, The
time trends shown in the figures were produced
using a Kernel smoothing technique. The Kernel
smoothing technique uses: a moving weighted-
mean smoother. The weighting function has a
maximum value at the center of the moving data
window and a value of zero at the edges of the
window.
Figure 3.3 shows time trends for S02 and par-
ticle SOj at seven CASTNet and six CAPMoN sites
for the period 1989 to 2002. The highest S02 and
SOj concentrations are observed in regions with
high S02 emissions (i.e., Indiana, Ohio,
Pennsylvania) while in contrast, the lowest con-
centrations occur in the northernmost and east-
ernmost regions of Canada, at sites distant from
major emission source areas. Consistent with the
large decline in eastern North American S02 emis-
sions during the 1990s:, all of the Canadian and
U.S. sites showed marked decreases in ambient
S02 and SOj concentrations between the early
and late 1990s. At most sites, the SOj and S02
trends lines follow each other closely with both
Figure; 3,2 - 98th percentile
PM j I concentrations at
Canadian TEOM and U.S. FRM
sites for the data years-
2000-2002,
(C an.'t'li.ir sil'-'s do n[ >I. all
include three full years
of data) :
species beginning their downward drop around
1989-91. At some sites (Vincennes, IN; Deer Creek,
OH; Prince Edward; VA), however, the decline in
SOj concentrations occurred two or three years
later than the decline in S02 concentrations. This
may be due to the close proximity of sources with
rapidly declining emissions, whereas particle SOj
concentrations may not decline as rapidly due to
relatively larger distances between the sources and
receptors. The S02 and SOj trends at Canadian
sites generally level off around 1998-2000 while
most U.S. sites continue a downward trend, with
SOj leveling off at only a few sites.
Particle NHJ .and total NOj concentration
trends are shown in Figure 3.4 for the same time
period, 1989 to 2002. Total NO^ is defined here as
the sum of gaseous HNOs and particle NO3 , both
reaction products of NOx. Figure 3.4 indicates that
particle NH+ concentrations in Canada remained
roughly constant throughout the period while U.S.
concentrations generally decreased between the
early and late 1990s., Ammonium concentrations
were considerably higher in the United States in
comparison to Canada, with the exception of the
site at Longwoods, which is located in a major
agricultural region of southwestern Ontario, a
large source of NH3 emissions.
Total NOg concentrations remained roughly
constant, throughout the 1989-2002 time period at
IO

-------
chapter 3
[micromoles/m3]
Ann Arbor
CAPMoN and CASTNet Air Concentrations
Figure 3 J - Long-term trends in the precursor-gases SO [green) and particulate SO^ (blue ) at rural CAPMoN and
CASTNet sites, 1989-2002,
1 1

-------
Canada - United States Transboundary PM Science Assessment
[micromoles/m3]
NHl HNOo+NO
Sutton , QU
HM03+ HO 3
HM03* M03
Vincormcs , IN
HM03+ N03
CAPMoN and CASTNet Air Concentrations
Figure 3.4 - Trends in total nitrate: {gaseous HNOs and particulate NOg) (green) and particulate NHJ (blue) at rural
CAPMoN and CASTNet sites, 1989-2002,
12


-------
Canada - United States Transboundary PM Science Assessment
B.C.
Alberta ;
Sas
Man
Ontario
c.
Quebec

N.B.
Nfld.











	i	
r	;	
 



 J
.% *  
 
  
 

 
  I
  i *
i
	Urban Sites
	Rural Sites

Figure 3.7 - The 98th percentile of Canadian 24-hour
PM, | concentratibns in 2QQ1. Sites shown are from
west to east. The Canada-Wi<3e Standard numerical
target of,30 yg/ffi is shown by the solid line. Data are
from continuous TEOM samplers.
Figure 3,8 - U.S regions used for data analysis
purposes.
generally experience higher PM2 3 concentrations
than do rural and remote sites. This pattern has
also; been observed in Alberta by Cheng et al.
(2000). However, rural sites can also experience
very high PM2 levels during large-scale PM
episodes,, often comparable to levels observed at
urban locations.
Figure 3.7 presents the one-year 98th per-
centile values of 24-hour PM2 5 concentrations in
2001 at monitoring sites that satisfied the 75 per-
cent NAPS data completeness criterion (or had a
98th percentile > 30pg/m3 as per the Canada-Wide
Standard Achievement document), shown by loca-
tion from west to east. In 2001, 98th percentile val-
ues were greater than 30 pg/m5 (shown by the red
line in Figure 3.7) at seventeen sites.. All of these
seventeen sites are in urban areas except for the
rural site of Simcoe, Ontario. Outside of Ontario
and Quebec, only Prince George recorded a 98th
percentile value greater than 30 pg/m3.
U.S. EPA's Air Quality System (AOS) as of April
2003, are presented here. PM2 5 data from the net-
work for Interagency Monitoring of Protected
Visual Environments (IMPROVE) are also
presented. Many data summaries are presented by
region, as shown in Figure 3.8, for understanding
potential differences in the characteristics of PM in
different parts of the United States, Four of these
regions border Canada.
Following the establishment of new ambient
standards for PM2 5 in 1997. the U.S. EPA led a
national effort to deploy and operate over 1000
PM2 5 monitors. The U.S. EPA has analyzed the
available data collected by this network from 2000-
2002. Data from the monitors were screened for
completeness with the purpose of avoiding sea-
sonal bias. To be included in these analyses, a
monitor needed to record at least a full year of
data, defined as either 4, 8, or 12 consecutive quar-
ters with eleven or more observations per quarter.
3.1.3 United States
The U.S. EPA and the states have been using a
national network to measure PM2 5 concentra-
tions since 1999. Summaries through the end of
2002, based on data publicly available from the
3.1.3.1 Spatial Variations in Annual Average
PM2 5 Concentrations across the United States
Figures 3.9 is a national map depicting county-
level annual mean PM2 5 concentrations from the
U.S. FRM network. The monitor with the highest
Northwest
California
Southwest
14

-------
chapter 3
Annual Means at U.S. FRM Sites, by Region (lower regions not shown)
	Whiskers display 90"
and 10th percentiles
	Dots identify
distribution means
	Completeness
criteria: 11 + samples
per quarter
00 '01 '02 '00 '01 '02 '00 '01 '02 '00 '01 '02 '00 '01 '02
yi regions (ri= 821) N orthwest (n=116) Uf-per Mid vest (n=72) Ind M ictoest (n=204) N ortheast (n=132)
PM Concentration (pgdn )
6 $3 couiflai
-or- I2
IS-:or- II
Figure 3:9 * County-leyel maximum annual mean
PM2 i concentrations, averaged over three years,
2000-2002.
Figure:!,10 - Annual PS%J means at U.S. FRM sites by
region over thres years, 2000-2002. The box identifies
the inter-quartile range, the ling in the middle
shows the median, whiskers display 90th and 10th
percentiles, and dots identify the distribution means.
concentration in each monitored county is used to
represent the value in that county. The map and
box plots show that many locations in the eastern
United States and in California had annual mean
PM2 5, concentrations above 15 ug/m3.
Annual mean PM2 5 concentrations were
above 18 ug/m3 in several urban areas throughout
the eastern United States, including Atlanta,
Birmingham. Chicago, Cincinnati, Cleveland.
Detroit, Indianapolis, Knoxville, Louisville,
Pittsburgh, and St, Louis. Los Angeles and the
central valley of California were also above
18 ug/m3,: Sites in the upper Midwest, Southwest,
and Northwest regions of the United States had
generally lower annual mean PM2 5 concentra-
tions, most below 12 ug/m3.
3.1.3.2 Annual Means of PM2 5 at U.S. FRM Sites
by Region
The annual PM2 5 mean concentrations across;
the northern regions of the United States range
from about 6 to 18 ug/m3, with a median of about
13 lig/m3. The 98th percentiles of the distribution
of 24-hour average concentrations range from
about 8 to 94 pg/m3, with a median of about
33 pg/m3. Figure 3.10 shows 3 years of annual
mean concentrations at FRM sites, for the data
years 2000-2002. Most FRM sites are urban
('Urban and Center City' or 'Suburban') according
to AQS definitions; FRM sites sample every day,
every 3rd day or every 6th day, with the predomi-
nant measurements being every 3rd day,
The left-most graph in Figure 3.10 shows the
three years of data for all sites in the United States
(irrespective of region) and the four other plots
show the northern U.S. regions bordering Canada.
PM2 5 concentrations decreased approximately
7 percent nationwide but the northern United
States did not see such a decrease. Except for the
Industrial Midwest, concentrations in the northern
regions have been much flatter. Average PM2 5
levels are lower than the U.S. averages in all north-
ern regions except for the Industrial Midwest
(Detroit. Cleveland).
3.1.3.3 Annual Means of PM2 5 at U.S. FRM Sites
within 300 km of Border by Region
Figure 3.11 focuses on U.S. FRM sites within 300
km of the Canadian border. This boundary was rec-
ommended based on various analyses of correla-
tion distance, back trajectories, and source attribu-
tion analysis. The left-most plot shows PM2 5 con-
15

-------
Canada - United States Transboundary PM Science Assessment
Annual Means at U.S. FRM Sites Within 300 km of the Canadian Border, by Region
	Whiskers display 90"
and 10th percentiles
 Dots identity
distribution means
	Completeness
criteria: 11 + samples
per quarter
'00 '01 '02 '00 "01 '02 '00 '01 '02 '00 '01 '02 '00 '01 '02
JI regions (n=158) Northviest (n=32) Upper Midwest (n=10) Ind. Mictoest (n=77) Northeast (n=38)
Figure 3.11 - Annual mean PM2 5 at U.S. FRM sites
within 300 km of the Canadian border by region, over
three years, 2000-2002.
Annual Means at U.S. IMPROVE Sites, by Region
(lower reaions not shown)
IT
Ii
 J,
I I

 Wtuskorsdis
play
90* and 10th
percentiles

 Dots identify

distribution means
 CompJeteness
criteria: It*

sampJaa pe< quarter
Al SntajSjrwi] MoriW^	U*f	W'H Ind	(n-ai hwmmi 1/1-il
Figure t.12- Annual PM2 5 means at rural U.S..
IMPROVE sites by region.
centrations at all U.S. sites within 300 km of the
border. This figure includes all of the sites in the 4
plots to the right (with one exception) since none
of the 'southern' regions have points that close.
The exception is one site (Alaska) which is not
included in a region, but meets the completeness
criterion. Mean PM2 5 concentrations for all sites,
(within 300 km) are relatively flat with the
Industrial Midwest driving the 'all regions' plot
since about half of the 158 sites are located there.
Sites in the Northwest show a large decline,
-22 percent (in average mean PM2 5 concentra-
tions) from 2000 to 2002. The 10 sites closest to
the Canadian border show a decline in mean
PM2 5 of 10 percent.
3.1.3.4 Annual Means of PM25 at U.S. IMPROVE
Sites by Region
Figure 3.12 shows the U.S. annual mean PM2 5 at
the rural IMPROVE network sites for the data years
2000-2002. PM2 5 levels are relatively unchanged
over the three years, with a slight increase in the
middle year (with the exception of the Northwest
region). Annual mean concentrations declined
from 1998 to 2001 at the three sites in the
Industrial Midwest. Annual mean levels of PM2 5
at sites in the Northwest and Upper Midwest are
consistent with national averages (at IMPROVE
sites), The levels in the two eastern regions, par-
ticularly the Industrial Midwest, are higher on aver-
age than the other sites.
3.1.3.5 Three year Annual Means and 98"'
Percentiles (2000-2002) of PM2 5 for U.S. Sites
(FRM) within 200 km of the Canadian Border
Figure 3.13 shows 3-year average 98th percentile
(triangle) and 3-year average annual mean (dot)
concentrations of PM2 5 at "border' sites. The data
for FRM sites are for the years 2000-2002. The dis-
tance criterion of 'within 200 km of the border' is
useful to show relationships, while removing any
significant clutter observed on the figures when
the distance from the border is increased. Sites
are shown (left to right) in a west-to-east longitude
order while the vertical lines separate the regions.
The first site (left-most) in the Northwest is really
located in Alaska (undefined region). The dashed
horizontal line at 15 |ig/m3 corresponds to the
annual U.S. National Ambient Air Quality Standard
for PM2 5. Numerous FRM sites in the Industrial
Midwest have annual means over the standard.
Only 1 site elsewhere (Northwest; Libby, Montana)
exceeds the annual standard. PM2 5 concentra-
tions measured at the IMPROVE sites (mean and
98th), while not displayed, are below most of the
concentrations measured at the FRM sites, as
expected from the rural and urban comparison.
16

-------
CHAPTER 3
60
45
30
15
	3-year annual mean dssplayed as dot 3-year average 98*1 percentile identified by star.
	\^stto East (longitude) order
	200k distance used m lieu of 300k to reduae datasel

A
M	^

A
A A
A A
A A 1 ^ O
CA	4
: ~ ~^"o *r v j
<-'Wj 
a\
'.V.A o*
o  i
Washington Montana Minn Michigan

Ohio


-------
chapter 3
iv5
Y el low knife, NW T errit*
E d m 01
ton
Quebe
peg ottaw1
Vancouver
ga-
Windsor
Victoria
S im coe
H a lifax
Hamilton
Pt. Petre Toronto
Figure j'J 6 - PM2  speciation data for NAPS network sites in Canada September 2001 -August 2002. Size of pie-graphs
indicates average PM,. s- concentration for the time period evaluated.
that episodic conditions at this rural site are
driven by secondary NO3 formation in the winter
season, and secondary SOj formation in the
summer. In addition to the differences observed
in PM2 5 composition between seasons, it is
suggested that there are major differences in PM
composition between urban and rural sites-.
Samples of PM2 5 from urban sites in Canada have
higher average fractions of black and organic
carbon and lower fractions Of SOj and NO3 than
rural sites. This is consistently attributed to the
increased contribution of the mobile source sector
(including on-road, off-road and diesel vehicles) in
urban areas.
3.2.1.T Chemical Composition of the Organic
Fraction of PM2 5
Of the organic mass that is chemically resolved in
measurements, it is estimated that primary carbon
is a larger component of the mass compared with
the products of VOC oxidation. To date, it is pos-
sible to identify only 10 to 20 percent of the organ-
ic species composing the total organic carbon frac-
tion of PM; however, monitoring technology for
this fraction is evolving. At present, measurement
information is insufficient for determining whether
the unresolved portion of the organic mass origi-
nates as direct organic particle emissions, VOC
emissions that condense directly to particles, p"
19


-------
chapter 3
CO

Regionalized Urban Speciation Patterns
Annual Average: Sep 2001-Aug 2002
25
20
15
3 10

3

S
O
I
&
 Crustal
~ TCM{h-1.4)
I Nitrate [H Ammoniium ~ Sulfate

H

lN=10l ra
z
:
|N=2|


M



H


H
r









	











u
Figure 3.18 -Annual average composition ofPM2 g in the United States by region (Urban data from the EPA Speciation
Trends Network).
The U.S. EPA speciation data in Figure 3.19
illustrate that sites in urban areas generally have
higher annual PM2 5 concentrations than the rural
stations shown in Figure 3.20. Urban sites in the
East include a large percentage of TCM, SOj, and
associated Nl I|, whereas., urban sites in the
Midwest and far West include a large percentage of
TCM and NOg. These patterns are also evident at
the Canadian locations (Figure 3.16). There are,
however, several sites in southern California where
the NOg fraction is of equal or greater proportion
than the carbon fraction.
The IMPROVE data in Figure 3.20 illustrate
that PM2 5 levels in the rural areas are highest in
the eastern United States and southern California,
as indicated by the larger circles. Sulphates and
associated NH4+dominate the east, with TCM as
the next most prevalent component. Sulphate
concentrations in the east largely result from S02
emissions from coal-fired power plants.- In
California and in the Midwest, TCM and NO^make
up most of the measured PM2 5.
Sulphates play a major role in the East,
Midwest, and South. Nitrates contribute to PM2 5
mass most in the Midwest and Northern locations,
Sites closest to the Canadian border (the North
Plains and Northwest sub-regions) are seen to
have relatively lower annual PM2 5 mass and con-
tain mostly carbon, SOj, and NO^, in that order.
For the domain of sites investigated, it is also seen
that the highest mass sites (for the year in ques-
tion) are in the East Coast, Northeast, and
Midwest.
Figure 3.21 shows seasonal variations for the
same grouping of urban and rural sites. In urban
areas, SOj and carbon dominate PM2 5 mass in
the summer season while NOj and TCM dominate
wintertime PM2 5 mass. Fall and spring show tran-
sitional amounts of each of the species when com-
pared to the summer and winter concentrations.
There is more NO^in the spring when compared to
the fall and higher TCM in the fall compared to the
spring.
2.1

-------
Canada - United States Transboundary PM Science Assessment
Ammonium
Nitrate
TCM
Crustal
18.69
31.18
J
Figure 3.19 * Summary of urban speciation data for PM25 in the United States ;(EPA Speciation Network). Size of pie
graphs indicates average PM, s Concentration for the time period evaluated.
A
Sulfate
6
Ammonium
&
Nitrate
A
TCM
*
Crustal
o
00
1.71
7.91 14,11
Figure 3.20 - Summary of rural speciation data (IMPROVE network). Size of pie graphs indicates average PM, ^
concentration for the time period evaluated.
22

-------
chapter 3
Select Urban Sites from EPA Speciation Network
Fall: Sep 2001-Nov 2001
sI 111
i s Si
g e
[BEtoaatiNoitteB
*"<581 * 55;!
I ! 11 ! I 11 f ;
z	 r 	
|Soit>e3s1| |"ert| |Nortt psi>s"l|Pes*rtWeita Norttweij
MASS ~ Ammonium El TCM (k=1.4)  Crustal
l~1 Sulfate  Nitrate
Select Urban Sites from EPA Speciation Network
Winter: Dec 2001 -Feb 2002
S 0.6
tf f 1 1
# * I I '
5 f S. c ]
 3. J s s s r a s | a a s | =  8 s - s b 8 g j
  T ti  "  15 ii Z  " " S" => = a * S S O 2  i-" :
2 " m	n i i 84"
MASS Q Ammonium H TCM (k=1.4)  Crustal
O Sulfate H Nitrate
CO I
CO I
CO
s 0.6
D_
5 a4
I 0.2
o
05
t 0
Select Urban Sites from EPA Speciation Network
Spring: Mar2002-May2002
I II II II II I III I III I III I II I I I I I I I I
ifli i9i!iRiPilili}jiiIilKi'li|iiiiili|i^spililiAi|iiiini|iliiiPi
I ! I I I M I I I M I I I ! I I : I l ! I M I M I I I
Iron
s ; ; S 2 s 2 2 s : : ; S : 8 2 ! s S 5  S s 2 ! i = 8 8 I
11!!! 1 f 1111 j f! 11 n i j I i 1! 11 i1! || j
I4IIJ ! " 2iSB)5S8i; s | "*
MASS ~ Ammonium^ TCM (k=1.4)H Crustal
f~l Sulfate  Nitrate
18

co
CO
1
16
F
s
OA
14
131
in

12


0.6
10
CC
IE
CL
O
0.4
6
cr
O
0.2
4


0
Select Urban Sites from EPA Speciation Network
Summer: Jun 2002-Aug 2002
I I I I laLIJaLI.IJ
iiiiBililih i iii ill pi
 r i ill ii ii
-------
Canada - United States Transboundary PM Science Assessment
I
20
15
10







Figure 3.22 - Spatial distribution of wet SO^
deposition (kg/ha/yr) in eastern North America,
1996-2001.
Figure 3.23 - Five-year (1996-2000) mean wet deposi-
tion exceedanceof critical SO4 loads (kg SOj/ha/yr)
for 95% lake protection level.
Ohio River Valley. When compared to the critical
loads for wet SOj deposition in eastern Canada1,
large areas of eastern Canada are receiving wet SOj
deposition in excess of critical loads (Figure 3.23).
There has been a decrease in observed lake
acidity near Sudbury, Ontario as a result of sub-
stantial reductions in S02 emissions from local
smelters and other sources outside the region.
However, in other areas of Ontario, Quebec and
Atlantic Canada, there has been a lack of change in
acidity and acid neutralizing capacity. This is partly
a result of the long-term depletion of base cations
in watershed soils, which control lake chemistry as
well as forest health. It is predicted that with cur-
rent emission reduction commitments., an area of
almost 800,000 km2 in southeastern Canada will
receive harmful levels of acid deposition in 2010.
Canada If currently using a geochemical
model, Model of Acidification of Groundwater in
Catchments (MAGIC), to analyze the current status
of lakes, rivers and forest soils and to predict
recovery timelines. The predicted response of lakes
and rivers to a hypothetical 50-percent S02 reduc-
tion scenario, despite a quick pH recovery, is a
base cation recovery lag time of 100 years (Clair et
aL 2003), The recovery period is predicted to be
much slower for forests.
3.3.2 Wet Nitrate Deposition
Nitrogen is a growth-limiting nutrient which is
taken up and retained by vegetation. However, in
many watersheds, prolonged NOj deposition has
resulted in soil acidification. It is possible that
even with reduced SOj deposition received by
ecosystems, the effects of continued NOg acidifica-
tion on forest and aquatic ecosystems will coun-
5-Year (I9%-20()0) Mean XS04 Wet Deposition Exceedance (kg/ha/vr)
*
1 Critical load values for wetSOy deposition to aquatic ecosystems in eastern Canada were estimated in 1990 (RMCC , 1990),. Values
were estimated using the- average geochemical characteristics of tertiary watersheds and assigning, a protection level for lakes f
95%. Areas with critical load values of lgss than 8 kg/ha/yr are considered to be very sensitive to acidification. It should he noted
however, that the use of wet SOJ deposition as; the primary environmental criterion for ecosystem protection has two limitations.
First, because the concurrent deposition of nitrate ions and base, cations has not been included, such a criterion considers'only1
residual SO^ deposition rather than the more general issue of residual acidification. The second limitation concerns the use.of wet
deposition information only. In eastern Canada, depending on the distance downwind from source regions, up to an additional 40%
of sulphur (and other chemical speciesj is dry deposited, contributing to acidification.
24

-------
chapter 3
5 ~
-------






-------
Canada - United States Transboundary PM Science Assessment
ambient concentrations of particle nitrate are
higher than summer concentrations. Summer and
winter NHa emissions for the 1996 base year are
shown in Figures 4.5a and 4.5b. Summer base-
case NH3 emission inputs for 2010 and 2020 are
provided in Figures 4.6a and 4,7a. Emissions of
ammonia are significant due to the role of ammo-
nia in the formation of ammonium sulphate and
ammonium nitrate.
The reduction in summer weekday S02 emis-
sions with the additional U.S. and Canadian con-
trols for 2010 and 2020 are shown in Figures 4.1c
and 4.2b. The reduction in winter weekday NOx
emissions with the additional U.S. and Canadian
controls are shown in Figures 4.3c and 4.4b. Only
reductions for these two PM precursors are shown
because the additional control measures for 2020,
discussed in Section 4.1.1.3, are concerned only
with these two pollutants (plus mercury for the
proposed Clear Skies legislation). Note that the
reductions in both S02 and NOx are concentrated
in the eastern half of the domain, which suggests
that the atmospheric response to these reductions
will also be concentrated in this region. Winter
base-case NH3 emission inputs for 2010 and 2020
are provided in Figures 4.6b and 4.7b. The emis-
sions of NH3 in the winter season are significant
because they are involved with the reaction of NOx
emissions to form particle ammonium nitrate.
Winter MEL emission inputs are significantly less
than summer NH3 emission inputs, particularly in
the U.S. portion of the domain.
Base Case S02 Emissions
0.000
Tons/Day
1996 Summer Weekday
m=199G_summer_so2.pave.new
800.00099

100.000
ZOO .000
100.000
50.000
20.000
10.000
5.000
2.000
1.000
January 1,0 0:00:00
Min= 0.000 at (1,1), Max= 937.948 at (106,56)
156
Figure 4.1a  1996
Summer weekday
S02 emissions for
Canada and the
United States;.
32

-------
chapter 4
15.000
2.000
1.000
u 0.000
tons/day
PAVE
MCfic
Min=
Hour: 00
0.000 at (1,1), Max= 773.069 at(106,56)
800.00099
400.000
200.000
100.000
50.000
20.000
10.000
Base Case S02 Emissions
2010 Summer Weekday
o=b10sum.ioapi
"J
Q-!_ ^ -
Figure 4. lb - 2010
Su mmer weekday
base case SQa
emissions for
Canada and the
United States.
Reduction in S02 Emissions from US/Canada Controls
2010 Summer Weekday
m=c10sum.ioapi, o=b10sum.ioapi
0.000 99
-1.000
-2.000
-5.000
-10.000
-20.000
-50.000
-100.000
-200.000
-400.000
u -800.000
tons/day
PAVE
by.
Hour: 00
Min=-656.482 at (106,56), Max= 48.351 at (101,40)
Figure 4.1c - 2010
Summer Weekday
reductions in $02
emissions for
Canada and the
United States.
33

-------
Canada - United States Transboundary PM Science Assessment
800.00099
400.000
Z00.000
100.000
50.000
20.000
10.000
5.000
2.000
1.000
0.000 1
tons/day
PAVE
MCifc
Base Case S02 Emissions
2020 Summer Weekday
c=b20sum.ioapi
Hour: 00
Min= 0.000 at(1,1), Max= 718.091 at(10S,56)
"i
v
Figure 4.2a - 202(3
Summer weekday
base case S02:-
emissiofis for
Canada arid the
United States,
Reduction in SO2 Emissions from US/Canada Controls
Summer Weekday (2020 Control -2020 Base)
i=c20sum.ioapi, k=b20sum.ioapi
0.000 99
-1.000
-2.000
-5.000
-10.000
-20.000
-50.000
-100.000
-200.000
-400.000
u -800.000 1
tons/day
1
156
by	nuur.uu
(CfiC	Min=-S37.967 at (106,56), Max= 31.439 at (89,63)
Figure 4.2b - 2020
Summer weekday
reductions in SCX,
emissions for
Canada and the
United States.
34

-------
chapter 4
Base Case NOx Emissions
1996 Winter Weekday
d=1996_wi nte r_n ox. p ave
500.00099
250.000
50.000
BO .000
40.000
20.000
10.000
5.000
2.000
1.000
0.000
January 1,0 0:00:00
Min= 0.000 at (1,1 J, Max= 709.046 at (89,60)
156
Figure 4.3a - 1996
Winter weekday
NO^.emissiens for
Canada and the
United States.
Base Case NOx Emissions
2010 Winter Weekday
h=b10wtr.ioapi
500.00099
J
250.000
150.000
80.000
40.000
20.000
10.000
5.000
2.000
1.000
0.000
tons/day
Hour: 00
Min= 0.000 at (1,1), Max= 547.459 at (89,60)
156
Figure 4.3b - 2010
Winter weekday
base case NOx
emissions for
Canada and the
United States.
35

-------
Canada - United States Transboundary PM Science Assessment
Reduction in NOx Emissions from US/Canada Controls
Winter Weekday (2010 Control -2010 Base)
f=c10wtr.ioapi, h=b10vrtr.ioapi
0.000 99
-0.500
1.000
-2.000
-4.000
B.000
15.000
30.000
50.000
-100.000
 -200.000 i
tons/day i
PAVE
M[$C
Hour: 00
Min=-214.775 at(106,5G), Max= 14.022 at(75,31)
156
Figure 4.3c - 2010
Winter weekday
reductions in NOx
emissions for
Canada and the
United States,
0.000 1
tons/day
PAVE
Mcfc
Base Case NOx Emissions
2020 Winter Weekday
d=b20wtr.ioapi
500.00099
250.000
150.000
BO .000
40.000
20.000
10.000
5.000
2.000
1.000
Hour: 00
Min= 0.000 at (1,1), Max= 472.384 at (89,60)
156
Figure 4.4a - 2020
Winter weekday
base case NOx
emissions for
Canada and the
United States.
36

-------
chapter 4
I
Reduction in NOx Emissions from US/Canada Controls
Winter Weekday (2020 Control - 2020 Base)
b=c20wtr.ioapi, d=b20wtr.ioapi
0.000 99
-0.500
-1.000
-2.000
-4.000
8.000
15.000
30.000
50.000
-100.000
^ -200.000 1
tons/day
PAVE
MCiSc
Hour: 00
Min=-217.544 at (106,56), Max=
4.680 at (62,77)
156
Figure 4.4b - 202D
Winter weekday
reductions in NO
emissions for
Canada and th
United State--
0.000 1
Tons/Day 1
Base Case NH3 Emissions
1996 Summer Weekday
e=199G_s u m me r_n h3. p ave
200.00099

100.000
50.000
20.000
10.000
5.000
2.000
1.000
156
January 1,0 0:00:00
Min= 0.000 at (1,1), Max= 140.621 at (83,25)
Figure 4.5a  1996
Summer weekday
N.H3. emissions for
Canada and the
United Statesv
37

-------
Canada - United States Transboundary PM Science Assessment
0.000 1
Tons/Day 1
Base Case NH3 Emissions
1996 Winter Weekday
f=1996_wi nte r_n h3. p ave
200.00099

100.000
50.000
20.000
10.000
5.000
2.000
1.000
January 1,0 0:00:00
Min= 0.000 at (1,1), Max= 139.731 at (83,25)
156
Figure 4.5b - 1996
Winter weekday
NHS emissions fe
Canada and the
United States.
0.000 1
tons/day
PAVE
Melt
Base Case NH3 Emissions
2010 Summer Weekday
h=b10sum.ioapi
200.00099
5
100.000
50.000
20.000
10.000
5.000
2.000
1.000
156
Hour: 00
Min= 0.000 at (1,1), Max= 165.425 at (83,25)
Figure 4.6a - 2010
Summer weekday
base case NH3
emissions for
Canadaand the
United States.
38

-------
chapter 4
0.000 1
tons/day
Base Case NH3 Emissions
2010 Winter Weekday
I=b10wtr.ioapi
200.00099
i
100.000
50.000
20.000
0.000
5.000
2.000
1.000
January 1,0 0:00:00
Min= 0.000 at (2,2), Max= 164.514 at (83,25)
156
Figure 4.6b - 20IS
Winter weekday
base case NHS
emissions for
Canada and the
United States.
LJ 0.000 1
tons/day
PAVE
b|
MCfic
Base Case NH3 Emissions
2020 Summer Weekday
d=b20sum.ioapi
200.00099
100.000
50.000
20.000
10.000
5.000
2.000
1.000
156
Hour: 00
Min= 0.000 at (1,1}, Max= 182.723 at (83,25)
Figure 4.7a -1020
Summer Weekday
base case Mils
emissions for
Canada and thfi
United States.
39

-------
Canada - United States Transboundary PM Science Assessment
^ 0.000 1
tons/day 1
PAVE
McSc
Base Case NH3 Emissions
2020 Winter Weekday
I=b20wtr.ioapi
200.00099
100.000
50.000
20.000
10.000
5.000
2.000
1.000
156
Hour: 00
Min= 0.000 at (1,1), Max= 181.767 at (83,25)
Figure 4.7b - 2020
Winter weekday
base case: NH-3
emissions lor
Canada and the
United States,
4.3 KEY SCIENCE MESSAGES
	Emission inventory information was combined
for both Canada and the United States to pro-
vide input to two multi-pollutant models
(AURAMS and REMSAD) for both base case and
control scenarios for the years 2010 and 2020.
	Emissions of S02 and NOx are projected to
decrease while NH3, VOCsand CO increase in
the future-year base cases. S02 and NOx emis-
sions are projected to decrease further with the
future-year control scenarios (2010 and 2020).
	Emissions of S02 and NOx under all consid-
ered scenarios are concentrated in the
Industrial Midwest Northeastern United States
and Southern Ontario, while emissions of NH3
are typically Concentrated further west in the
central Midwest region.
The emissions of S0Ol NO
and NH3, and their
contributions to PM2 5 levels vary seasonally.
NH3 emissions and biogenic NOx (and VOC)
emissions have the largest seasonal variations.
40


-------
Canada - United States Transboundary PM Science Assessment
were: (1) 1996 base case, (2) 2010 base case, (3)
2020 base case, (4) 2010 control case, and (5) 2020
control case. The 1996 base case results were used
to evaluate the performance Of REM SAD in pre-
dicting observed concentrations in 1996. The
results of this model performance evaluation are
provided in the Appendix. Existing controls (i.e.
legislation/ agreements) in each country were
included in the 2010 and 2020 base case runs while
the 2010 and 2020 control runs contain additional
anticipated controls for each country (as described
in Chapter 4). The REMSAD model was used to
estimate hourly air-quality concentrations and
acid deposition for an entire year for each model
run. This section reports results for the 2010
annual PM2 5 concentration and the 2020 annual,
January, and July PM2 5, SOJ, N03", and NHJ
concentrations. The modelling domain used in
REMSAD is shown in Figure 5.1.
5.1.1 REMSAD Results
Annual average ambient PM2 5 concentrations for
the 2010 base case are provided in Figure 5.2a
while annual. lanuary, and July average PM2 5 con-
centrations for the 2020 base case are provided in
Figures 5.3a, 5.4a, and 5.5a. Annual, lanuary, and
July average particle SOj concentrations for the
2020 base case are provided in Figures 5.6a, 5.7a,
and 5.8a and annual, January, and July average par-
ticle NOgConcentrations for the 2020 base case are
shown in Figures 5.9a, 5.10a, and 5.11a. Annual,
January, and July average NHJ concentrations for
the 2020 base case are illustrated in Figures 5.12a,
5.13a, and 5.14a. Figure 5.2b shows annual average
PM2 5 air quality concentration reductions for
2010 that result from the implementation of con-
trols in the United States and Canada. Annual,
January, and July average PM2 5 concentration
reductions for the 2020 scenario are provided in
Figures 5.3b, 5.4b, and 5.5b. Annual, January, and
July average SOj concentration reductions that
result from U.S. and Canadian controls in 2020 are
shown in Figures 5.6b, 5.7b and 5.8b while Figures
5.9b, 5.10b, and 5.11b illustrate annual, January,
and July average particle NO~ concentration reduc-
tions that result from U.S. and Canadian controls
in 2020. Annual, January, and July average NHJ
concentration reductions that result from the 2020
controls are shown in Figures 5.12b, 5.13b and
5.14b. Many1 control measures already underway
Figure 5.1 - REMSAD modelling
domain (-36x36 km2), Grid
squares eneampass 1/2 degree:
longitude, 1/3 degree latitudes
g-W range; 54 degrees W - 13?
degrees W; N^S range:
22 degrees N - 55 degrees N.
Vertical extent: Ground to
16,200 meters flOOmb) with
12 layers,
42


-------
Canada - United States Transboundary PM Science Assessment
Figure 5.2a - Annual
average- PM, !
concentrations
2010 baBe-casei.
Figure 5.2b -
Reductions in
annual PMj g
cjincentrations
from controls
in 2010
PM2.5 Concentrations
LJ 0.000 1
ug/m3
PAVE
tcrlc
Annual 2010 Base Case
b=csa10be1p1_v706ext.ioapi
24.000 98
21.000
18.000
15.000
2.000
9.000
6.000
3.000
a
Hour: 00
Min= 0.298 at (155,2), Max= 51.001 at (131,93)
155
Reduction in PM2.5 Concentrations from US/Canada Controls
Annual Concentrations (2010 Control - 2010 Base)
b=csa10be1p1_v706ext.ioapi, c=csa10ce1 p1_v706ext.ioapi
-0.000 98
-0.300
-0.600
-0.900
1.200
-1.500
-1.800
2.100
-2.400
ug/m3
155
Hour: 00
Min= -1.818 at (105,55), Max= 0.005 at (18,45)
44

-------
chapter 5
PM2.5 Concentrations
18.000
15.000
12.000
2020 Annual Base Case
d=csa20be1p1_v706ext.ioapi
24.000 98
21.000
9.000
6.000
3.000
0.000
ug/m3
PAVE
MCifc
Min=

Hour: 00
0.295 at (155,2), Max= 54.882 at (131,93)
Figure 5.3a - Annual
average PM,. e
concentrations
2020 base easg-
Reduction in PM2.5 Concentrations from US/Canada Controls
Annual Concentrations (2020 Control - 2020 Base)
d=csa20be1 p1_v706ext.ioapi, e=csa20ce1p1_v70Gext.ioapi
-0.000 98
-0.300
-0.600
-0.900
-1.200
-1.500
-1.800
-2.100
-2.400
ug/m3
PAVE
by.
Hour: 00
Min= -2.301 at(112,55), Max= 0.001 at(2S,52)
Figure 5.3b -
Reductions in
annual PMj j
Concentrations
from controls
in 2020.
45

-------
Canada - United States Transboundary PM Science Assessment
LJ 0.000 1
ug/m3
PftVE
Mcfc
PM25 Concentrations
January 2020 Base Case
s=csa20be1 pi v706ext.01 .monavg.dat.ioapi
24.000 98
21.000
18.000
15.000
12.000
9.000
6.000
3.000
S
155
Hour: 00
Min= 0.196 at (155,2), Max= 50.980 at (117,71)
Figure 5..4a. -
January average
PM2 s concentra
tions n base
case.
Reduction in PM2.5 Concentrations with US/Canada Controls
January 2020 (Control - Base)
q=csa20ce1p1_v706ext.01.monavg.dat.ioapi, s=csa20be1 pi_v706ext.01 .monavg.dat.ioapi
-0.000 98
-0.300
-0.600
0.900
-1.200
-1.500
-1.800
-2.100
-2.400
ug/m3
155
Hour: 00
Min= -1.858 at(101,35), Max= 0.011 at(27,42)
Figure 5,4b -
ReduCtiotts in
January PM2j
conggntrations
from controls
'in
46

-------
chapter 5
PM2.5 Concentrations
18.000
15.000
12.000
9.000
6.000
3.000
pi 24.000 98
21.000
July 2020 Base Case
t=csa20be1p1_v706ext.07.moriavg.dat.ioapi
0.000
ug/m3
PAVE
tele
1
1	155
Hour: 00
Min= 0.194 at (155,2), Max= 46.473 at (131,93)
Figure 5.5a - July
average PML, a
concentrations
2020 base case.
Figure 5.5b -
Reductions in
July fiSfcjj
concentrations
from Controls
in 2020.
-1.800
Reduction in PM2.5 Concentrations with US/Canada Controls
July 2020 (Control - Base)
r=csa20ce1 p1_v706ext.07.monavg.dat.ioapi, t=csa20be1p1_v706ext.07.monavg.dat.ioapi
-0.000 98
I -1.200
' -1.500
-2.100
-2.400
ug/m3
PAVE
1

Hour: 00
Min= -3.308 at (111,52), Max= 0.010 at (18,76)
47

-------
Canada - United States Transboundary PM Science Assessment
Sulfate Concentrations
3.600
3.000
2.400
n 4.800 98
4.200
2020 Annual Base Case
=csa20be1 pi _v706ext.ioapi
Hour: 00
MCflu	Min= 0.123 at (155,2), Max= 4.870 at (103,56)
d
1.800
1.200
0.600
u 0.000 i
ug/m3
V
IC'V"'-*.
Figure 5.6a - Annual
average: SOy
Concentrations
2020 base: case.
Reduction in Sulfate Concentrations from US/Canada Controls
Annual Concentrations (2020 Control - 2020 Base)
d=csa20be1 p1_v706ext.ioapi, e=csa20ce1p1_v706ext.ioapi
0.00 98
-0.20
-0.60
-0.80
-1.00
-1.20

-1.S0
ug/m3
PAVE
Hour: 00
Min= -1.41 at (106,56), Max= 0.02 at (48,45)
Figure 5.6b -
Reductions in
annual SO^
concentrations
from Controls
in 2020,
48

-------
chapter 5
0.000 1
ug/m3
PAVE
Mcifc
Sulfate Concentrations
January 2020 Base Case
s=csa20be1 p1 v706ext.01 .monavg.dat.ioapi
4.800 98
4.200
3.600
3.000
2.400
1.800
.200
0.600
Hour: 00
Min= 0.045 at (155,2), Max= 2.282 at (108,32)
155
Figure 5.7a - January
average SOj
concentrations
2020 base case.
Reduction in Sulfate Concentrations from US/Canada Controls
January 2020 (Control - Base)
q=csa20ce1p1_v706ext.01.monavg.dat.ioapi, s=csa20be1p1 v706ext.01.monavg.dat.ioapi
0.000 98
0.200
-0.400
-0.600
-0.800
-1.000
1.200
-1.400
-1.000
ug/m3
Hour: 00
Min= -0.558 at (129,35), Max= 0.005 at (48,45)
155
Figure 5.7b -
Reductions in
January SOj
concentrations
from controls
in 2020.
49

-------
Canada - United States Transboundary PM Science Assessment


0.000
ug/m3
PAVE
b!
Mcifc
Sulfate Concentrations
July 2020 Base Case
t=csa20be1p1 v706ext.07.monavg.dat.ioapi
4.800
4.200
3.600
3.000
2.400
1.800
1.200
0.600
155
Hour: 00
Min= 0.045 at (155,2), Max= 6.975 at (114,54)
Figure 5.8a - July
average S O j
oonsJSntrations
2020 base case.
Reduction in Sulfate Concentrations with US/Canada Controls
July 2020 (Control - Base)
r=csa20cel pi _v70Gext.07.monavg.dat.ioapi, t=csa20be1 p1_v706ext.07.monavg.dat.ioapi
0.000 98
-0.200
-0.400
-0.600
-0.800
1.000
1.200
1.400
1.600
ug/m3
155
Hour: 00
Min= -2.436 at (111,50), Max= 0.002 at (19,76)
Figure 5,8b .
Reductions in
JulySOj
cont^nttations
from controls
in 2020.
50

-------
chapter 5
fl

^ 0.000 1
ug/in3
PAVE
Mclc
Particle Nitrate Concentrations
2020 Annual Base Case
d=csa20be1 p1 _v706ext.ioapi
4.800 98
4.200
3.600
3.000
2.400
1.800
1.200
0.G00
155
Hour: 00
Min= 0.000 at (52,6), Max= 4.163 at (112,55)
Figure 5.9a - Annual
average NO^
concentrations
2020 base case.
Change in Nitrate Concentrations with US/Canada Controls
Annual Concentrations (2020 Control - 2020 Base)
d=csa20be1p1_v706ext.ioapi, e=csa20ce1p1_v70Gext.ioapi
I

0.150 98
0.050
0.050
0.150
0.250
0.350
0.450
0.550
0.650 1
ug/m3
PAVE
Hcfc
155
Hour: 00
Min= -0.640 at (90,53), Max= 0.089 at (121,57)
Figure 5.9b -
Reductions in
annual NO^,
concentrations
from controls
in 2020.
51

-------
Canada - United States Transboundary PM Science Assessment
n
0.000 1
ug/m3
PAVE
mcNc
Particle Nitrate Concentrations
January 2020 Base Case
s=csa20be1 p1 v706ext.01 .monavg.dat.ioapi
4.800 98
4.200
3.600
3.000
2.400
.800
1.200
0.600
s.
Hour: 00
Min= 0.000 at (154,3), Max= 5.925 at (101,35)
155
Figure 5.1.0a -
January average
NO J concentrations
2Q20 base case.
Change in Nitrate from US/Canada Controls
January 2020 (Control - Base)
q=csa20ce1p1_v706ext.01.monavg.dat.ioapi, s=csa20be1p1_v706ext.01.monavg.dat.ioapi
0.150 98

s
%
0.050
0.050
-0.150
-0.250
-0.350
0.450
-0.550
-0.650 1
ug/m3
PAVE
MCifc
Hour: 00
Min= -1.093 at (101,35), Max= 0.102 at (118,49)
155
Figure 5.1Db -
Reductions in
January N03
concentrations
from controls
in 2020.
52

-------
chapter 5


0.000 1
ug/m3
PAVE
MClfc
Particle Nitrate Concentrations
July 2020 Base Case
t=csa20be1 pi _v706ext.07.monavg.dat.ioapi
4.800 93
4.200
3.600
3.000
2.400
1.800
1.200
0.600
155
Hour: 00
Min= 0.000 at (153,4), Max= 2.375 at (29,36)
Figure 5.1 la - July
average 80J
concentrations
2020 base case,
Change in Nitrate Concentrations with US/Canada Controls
July 2020 (Control - Base)
r=csa20ce1p1_v706ext.07.monavg.dat.ioapi, t=csa20be1p1_v706ext.07.monavg.dat.ioapi
0.150 98
5

0.050
0.050
0.150
0.250
0.350
0.450
0.550

-0.650 1
ug/m3
PAVE
MCijc
Hour: 00
Min= -0.402 at(106,66), Max= 0.191 at(117,60)
155
Figure 5.1 lb -
Reductions in
July N03
concentrations
frorn controls
in 2020.
53

-------
Canada - United States Transboundary PM Science Assessment
Ammonium Concentrations
1.800
1.500
1.200
0.900
0.600
0.300
Hi 2.400 98
2.100
2020 Annual Base Case
d=csa20be1 p1_v706ext.ioapi
0.000 1
ug/m3
PAVE
mcNc
Hour: 00
Min= 0.028 at (155,2), Max= 2.756 at (114,54)
i
	2_l
Figurfe5.12a-
Annual average
NHJ concentrations
2020 base case,
Figure 5.12b -
Reductions in
annual NHJ
concentrations
from controls
in 2020.
Reduction in Ammonium Concentrations from US/Canada Controls
1-0.10
-0.20
-0.30
-0.40
-0.50
-0.60
Annual Concentrations (2020 Control - 2020 Base)
d=csa20be1p1_v706ext.ioapi, e=csa20ce1p1_v706ext.ioapi
0.00 88
-0.70
u -0.80
ug/m3
PAVE
by.
1
Hour: 00
Min= -0.58 at (112,55), Max= 0.00 at (46,19)
vv
v
54

-------
chapter 5
0

0.000 1
ug/m3
PftVE
MCffc
Ammonium Concentrations
January 2020 Base Case
s=csa20be1p1_v70Gext.01.monavg.dat.ioapi
2.400 98
2.100
1.800
1.500
1.200
0.900
0.600
0.300
155
Hour: 00
Min= 0.017 at (155,2), Max= 2.429 at (101,35)
Figure 5,13a -
January average
NHJ concentrations
2020-base case.
Reduction in Ammonium Concentrations with US/Canada Controls
January 2020 (Control - Base)
q=csa20ce1p1 _v706ext.01 .monavg.dat.ioapi, s=csa20be1p1_v706ext.01.monavg.dat.ioapi

o.ioo
-0.200
-0.300
-0.400
-0.500
-0.600
-o./oo
m -0.800 1
ug/m3
PAVE
MC&
Hour: 00
Min= -0.446 at (101,35), Max= 0.003 at (27,42)
Figure 5.13b -
Reductions in
January NHJ
concentrations
from controls
in 2020.
55

-------
Canada - United States Transboundary PM Science Assessment
a

0.000 1
ug/m3
PftVE
MC^C
Ammonium Concentrations
July 2020 Base Case
t=csa20be1 pi v706ext.07.monavg.dat.ioapi
2.400 98
3
Z.I 00
1.800
1.500
1.200
0.900
0.600
0.300
S
Hour: 00
Min= 0.017 at (155,2), Max= 2.930 at (114,54)
155
Figure 5.14a -
July average NHJ"
eortcentrations 2020-
base case.
Reduction in Ammonium Concentrations with US/Canada Controls
July 2020 (Control - Base)
r=csa20ce1p1_v706ext.07.monavg.dat.ioapi, t=csa20be1p1_v70Sext.07.monavg.dat.ioapi

0.000 98
0.100
0.200
0.300
0.400
0.500
0.600
-0.700
0.800 1
ug/m3
PAVE
M(Sc
Hour: 00
Min= -0.848 at (112,55), Max= 0.035 at (143,35)
Figure 5.14b -
Reductions in
July NHJ
concentrations
from controls
in 2020.
56





-------
chapter 5
Figure 5.16 - Nine-daiy-average PM2 5 mass Concentration field and PM2 5 inorganic chemical component aDrteentration
fields predicted by AURAMS for the Pgb. 7-15, 1998 winter period for the "2010 base" case-ernissions: (a) top left panel -
PM2 s mass; (b) top right panel - Pf|3 5 SOJ mass; (cj lower left panel - PMj j NHjmass; :(d): lower right panel -
PM2 S: N03 mass. All fields are at 15 m height in units of jig/m%
61

-------
Canada - United States Transboundary PM Science Assessment
Figure 5.17 Nine-day-average PM 5 mass concentration difference field and PM2 5 inorganic chemical component
concentration difference fields predicted byAURAMS for the Feb. 7-15, 1998 winter period for the "201Q control" ease
minus the "2010 base" esse: (a) top left panel - PM2 -s mass; (b) top right panel - PMa  SOj mass; (C); lower left panel
- PM2 5 NHJ mass-, (d); lower right panel - PM2 5 N03mass. All fields are at 15 m height in units: of |.ig/m3. Negative
values denote a reduction for the "2010 control" case relative to the "201Q base" casei
62

-------
chapter 5
ug/m3
Min = 0.00 at (58.94N, 73.01 W) Max = 34.01 at (41.69N, 87.86W)
Figure 5.18 - Same as Figure 5.16 but for the "2020 base" case emissions.
63

-------
Canada - United States Transboundary PM Science Assessment
Figure 5.19 - Same as Figure 5.17 but for the "2020 control" case minus the "2020 base"
PM2.5
ug/m3
0.0
-1.0
-2.0
-3.0
-4.0
-5.0
-6.0
-20.0 I
PM2.5
N03
ug/m3
PM2.5
S04
ug/m3
0.0
Min = -2.42 at (41.60N, 92.46W) Max = 1.32 at (36.74N, 86.68W)
Min = -0.55 at (41.60N, 92.46W) Max = 0.23 at (36.74N, 86.68W)
Min = -1.46 at (31.88N, 83.87W) Max = 0.50 at (42.89N, 95.96W)
Min = -1.89 at (41.65N, 92.90W)
Max = 1.57 at (36.74N, 86.68W)
64



-------
chapter 5
Figure 5 20 -: ElgSien-day-aVefage PM2-5 mass Concerttration field and PUL 5 inorganic chemical component
Concentration fields predicted by AURAMS for the July S-18, 1995 summer period for the "2010 base" rase emissions:
(a) top left panel - JWJB mass; (a) top right panel - PM j5 SOj mass; (c) lower left panel - PM, 5 NHJ mass; (d) lower
right panel - PM2 s Nt|g mass. All fields are at 15 m height in units of |Jg/m3.
S04
ug/rn3 I
	
Min = 0.00 at (60.57N, 83.03W) Max = 36.18 at (38.99W, 81.57W)
PM2.5
N03
ug/m3 I
17.5 |
Min = 0.00 at (24.92N, 80.36W) Max = 16.78 at (30.23N, 89.84W)
67

-------
Canada - United States Transboundary PM Science Assessment
PM2.5
ug/m3
35.0 |
Min = 0.17 at (63.16N, 93.45W) Max = 49.57 at (40.93N, 73.97W)
PM2.5
S04
ug/m3 I
17.5 |
Min = 0.00 at (60.57N, 83.03W) Max = 8.27 at (30.23N, 89.84W)	Min = 0.00 at (24.92N, 80.36W) Max = 16.78 at (30.23N, 89.84W)
Figure 5.21 -: Eleven-day-average PM2 5 mass concentration difference field and PM2 5 inorganic chemical component
concentration difference fields predicted by AURAMS for the July 8-18, 1995 summer period for the "2010 control" case
minus the "2010 base" case: (a) top left panel - PM2 5 mass; (b) top right panel - PM2 5 SO^ mass; (c) lower left panel -
PM2 5 NHJ mass; (d) lower right panel - PM2 5 NO3 mass. All fields are at 15 m height in units of jag/m3. Negative values
denote a reduction for the "2010 control" case relative to the "2010 base" case.
.a	
Min = 0.00 at (60.57N, 83.03W) Max = 36.18 at (38.99N, 81.57W)
68

-------
chapter 5
PM2.5
S04
ug/m3
Min = 0.17 at (63.16N, 93.45W) Max = 49.63 at (40.93N, 73.97W)	Min = 0.00 at (60.57N, 83.03W) Max = 33.86 at (34.33N, 84.60W)
PM2.5
N03
ug/m3
Min = 0.00 at (60.57N, 83.03W) Max = 8.15 at (30.23N, 89.84W)	Min = 0.00 at (24.52N, 79.17W) Max = 15.83 at (30.23N, 89.84W)
Figure 5,22 - Same, as Figure 5,20 but lor M "2020 base" case emissions.
69

-------
Canada - United States Transboundary PM Science Assessment
PM2.5
NH4
ug/m3
0.0
Min = -18.37 at (34.33N, 84.60W) Max = 1.64 at (27.91 N, 96.74W)
Min = -2.51 at (33.SON, 81.68W) Max = 0.18 at (34.50N, 78.03W)
Min = -19.01 at (34.33N, 84.60W) Max = 0.07 at(45.65N, 96.17W)
Max = 3.81 at (43.24N, 77.56W)
PM2.5
N03
ug/m3
= -1.04 at (33.08N, 89.98W)
Figure 5.23 < Same as Figure 5.21 but for th*2020 control" case minus the "2020 base" case.
70


-------
Canada - United States Transboundary PM Science Assessment
5.3 RESULTS OF CMAQ
MODELLING IN THE
Georgia Basin - Puget
Sound Region
The Community Multiscale Air Quality (EPA, 1999;
MCNC, 2001) modelling system was applied over
the Pacific Northwest to gain insight into the sig-
nificance of the transboundary transport of air pol-
lutants across- the international border separating
British Columbia and Washington State, and to
determine the impacts of forecast changes in pol-
lutant emissions expected by 2010 and 2020 on
ambient air quality in 2000.
The version of CMAQ used for this work is the
June 2001 version that was parallelised (RWDI,
2003a) for a PC/Linux cluster running Redhat Linux
v7.3. The photochemical mechanism used was the
*radm2_ae2_aq' mechanism. This mechanism was
selected in order to; be compatible with CMAQ
modelling being performed by others over the
Pacific Northwest. The CMAQ modelling domain
used encompasses the Pacific Northwest stretch-
ing from central Oregon to central British
Columbia and from western Idaho to the Pacific
Ocean, or in Other words an 800 km wide domain
straddling 500 km each side of the Canada/US
border with a domain resolution of 12 km. Nested
within this domain is a 4-km fine resolution sub-
domain centred over the Georgia Basin and Puget
Sound. Resolutions of this magnitude are
required in order to try to account for the complex
terrain and marine environments of the Pacific
Northwest. See Figure 5.24 for geographical refer-
ences and domain extents.
In this case, the CMAQ chemistry transport
model is driven using the MC2 (Mesoscale
Compressible Community) meteorological model.
MC2 is based on the Euler equations and is a fully
compressible non-hydrostatic model using gener-
alised terrain-following coordinates. Complete
descriptions of MC2 are available in Laprise
(1997). The MC2 meteorology is at a resolution of
3,3 km using version 4.9.1 of the MC2 dynamics


CMAQ 12km Domain

British

Columbia

f
	"CMAQ 4km Domain^


%, % M
% % s
Vancouver FVRD 3 '
n #GVRD a ^
& JJncan Whatcom County
V,
-------
chapter 5
and version 3.7 of the RPN/CMC physics package.
This meteorology was then interpolated and repro-
jected onto the CMAQ grid and remapped for
ingest by meteorology/chemistry interface proces-
sor (MCIP) of CMAQ and for ingest by the SMOKE
emissions model.
The CMAQ simulations were performed using
meteorology for a typical summer period and for a
typical winter period. The summer period selected
was August 09-20, 2001. This period embraced a
dry blocking weather pattern of two regimes; a
stagnant phase, and a well-mixed phase. This
period coincided with the Pacific 2001 Field Study
(Li, 2001) from which there was a rich meteorolog-
ical and chemical dataset. The winter period
selected was December 01-13, 2002. This period
comprised a short stagnant phase, followed by a
weak blocking pattern, and ended with a transient,
well-mixed phase. For both summer and winter
periods, 2000 emissions inventory data were used.
There were no known significant anthropogenic
emission differences between 2000,2001, and 2002.
The Pacific 2001 field study dataset was used
to evaluate CMAQ performance over the Georgia
Basin-Puget Sound region. In spite of the complex
terrain and marine environment challenges of the
Pacific Northwest, it was felt that the CMAQ per-
formance was consistent with that found by others
in Canada and the United States. Overall, the
model performed well for predicting PM2 5 at both
the 12-km resolution and 4-km resolution
domains. It should be noted that the CMAQ "1+1"
particle mass was used as if it were PM2 5, even
though the difference can be substantial both con-
ceptually and quantitatively. Subsequent refer-
ences will not make this distinction.
Overall, the diurnal patterns and magnitude of
the modelled daily average PM2 5 levels were quite
good. In general, the 4-km PM2 5 results were bet-
ter than those for the 12-km domain, particularly at
night. This is believed to be the results of local
emission sources and the more heterogeneous
nature of PM2 5 as a regional pollutant compared to
ozone. Secondary particulate matter can form very
rapidly or slowly depending on the environmental
conditions and emission source characteristics.
5.3.1 Qualitative Analysis of Simulations
for the 2000 Base Case
5.3.1.1	Summer PM2 5
In the 12-km grid domain, PM2 5 starts to build up
in the vicinity of the major primary sources
(urban/industrial/marine areas) after about 24
hours of model 'spin-up'. The combination of sea
breeze and an onshore westerly flow pushes the
PM2 5 concentrations inland, toward the east away
from the urban and marine areas during the day-
time. And, mountain flows from the northeast
along the Fraser Valley push the pollutants back
toward the west during the night. This day-night
pattern in PM2 5, levels persists until the onset of
the well-mixed phase. In the 4-km simulations,
results are similar but show somewhat improved
resolution of focal hotspots near the sources of
primary PM2 5 emissions (Figure 5.25),
5.3.1.2	Winter PM2 5
During the model 'spin-up' and stagnant meteoro-
logical periods (December 01-07), PM2 5 levels
build up around and slightly downwind (east) of
the urbanized areas of the Greater Vancouver
Regional District (GVRD), Seattle, and Portland.
During the weak blocking period (December 07-
10), offshore flows and land breeze effects push the
urban plumes toward the west and over the Pacific
PM2J
PM 2.5 Concentration
Au^ut" 200 ' Com 4 hi
Figure 5.25 - PM2,5 Concentrations for the. August,
2601 summer base ease, predicted oyer the CMAQ
domain on a 4x4 km2 grid.
73




-------
chapter 5
Ozone
Max = 8.6 at (4Z.02N, 87.24W)
Min = -16.9 at(39.38N, 81.35W)
Figure 5.2<5 - Peak ozone concentration
difference, field at 15 m height for the
July 12-15, 1995 summer period for' the
"2020 Control" case minus the "2020 base"
case. The two peak ozone concentration
fields were constructed by averaging over
the afternoon period (15 - 21 UTC) for
4 days (July 12th to 15th) corresponding
to a regional Ozone episode.
Figure 5.27 -Annual
reduction in SCgj wet
deposition from additional
U.S,_and Canadian controls
(2020 control vs. base.]:.
Reduction in Sulfate Wet Deposition from US/Canada Controls
Annual Wet Deposition (2020 Control - 2020 Base)
O=csa20ce1p1_v706ext.2020.wet.yrsum.dat.ioapi, u=csa20be1 p1 v706ext.2020.wet.yrsum.dat.ioapi
0.00 98
. -0.75
I -1.50
-2.25
-3.00
-3.75
-4.50
-6.00
v
Hour: 00
Min= -6.66 at (101,55), Max- 0.00 at (49,6)
77

-------
Canada - United States Transboundary PM Science Assessment
Reduction in Nitrate Wet Deposition from US/Canada Controls
Annual Wet Deposition (2020 Control - 2020 Base)
O=csa20ce1p1_v706ext.2020.wet.yrsum.dat.ioapi, u=csa20be1 p1_v706ext.2020.wet.yrsum.dat.ioapi
0.00 98
-0.50
-1.00
-1.50
-2.00
-2.50
-3.00
-3.50
-4.00
kg/ha
PAVE
Hour: 00
Min= -4.45 at (110,66), Max= 0.02 at (19,74)
Eigure5.28 -
Annual reduction
in NOjwet
deposition from
additional O.S!
and Canadian
controls (2020:
control fe. base).
Eigure:5.29 -
Aerosol light
extinction
(in Mm"1) for the
haziest- 2-G percent
days and. contribu-
tion by individual
particulate matter
constituents, based
on 1997-1999
IMPRCS/E data
(USEPA, 1999).
Species
Sulfate
Nitrate
Organic Carbon
Elemental Carbon
Crustal Matter
78


-------


-------
Canada - United States Transboundary PM Science Assessment
These results can be thought of as an indica-
tor combining the upwind status of a
state/province, the geographic size of the
state/province, and the magnitude of source emis-
sions within the state/province. A state or
province that is close to, and frequently upwind of.
multiple Class 1 areas- will generally contribute
more mass than states or provinces that are sel-
dom upwind, unless the concentration difference
is marked. For example, Minnesota contributes a
large percentage of mass to Boundary Waters (35.2
percent) although the average concentration asso-
ciated with air masses in Minnesota is less than 6
|ig/m3:,. Similarly the Canadian provinces make
significant contributions to the border-area Class 1
sites; Ontario provides about 16 percent of the
annual PM2 5 mags at Boundary Waters and
Quebec provides about 18 percent to Acadia. Ohio
and Pennsylvania are associated with high-con-
centration air masses at the three Class 1 sites
shown, but only make significant (>5 percent) con-
tributions to annual PM2 5 mass at the nearby
Dolly Sods Wilderness site.
In an exactly analogous manner, the contribu-
tion of each state and province to the joint set of
17 Class 1 .areas was derived (not shown),. The
results indicate that seme states associated with
high-concentration air masses nevertheless con-
tribute only a small amount of mass to the collec-
tive group of Class 1 sites? conversely, states (or
provinces) with low average concentrations can be
major mass contributors.
Average RM
I | 46.5.6
56-7 1
93.119
113-137
13 7 14 0
Average OC
I I 09
14-1.5
53-Jfi
Averago S04
I I '3-18
I I '6-?
2 7-34
34 -4 A
i .1 -1 y
>9-5?
Average EC
j j u .
03-03
D 3 - 0 4
04-05
M 0 8
05-08
Average NO3
] 0.2-0 3
03 0 4
04-05
05-0
05-08
08 11
Average Sail
Q2 - 03
03-03
0 3-04
ii.l l)^
n 5 - n 
9 6 1 1
Figure 6.1 - Average concentrations of PM2 5 and components (pg/m3) by state and province (IMPROVE sites shown
as blue dots). Each trajectory endpoint is associated with concentrations corresponding to the IMPROVE sample for
the trajectory start date.
82


-------
Canada - United States Transboundary PM Science Assessment
Figure 6.2 - Urban Excess
Analysis for SO^, NHJ, NO3,
TCM and crustal material for
13 urban areas in the United
States
(Note: k=l .8 in order to convert
carbonaceous mass into TCM).
CLO 0 4 0.9
0.0 0.9 1.9
Bronx
Fresno
Tulsa
Altenta
23 8.1 13.2
6.1.1.3 Sources of PM2 5 to Eastern North America
An ensemble-trajectory analysis technique known
as Quantitative Transport Bias Analysis (QTBA;
Keeler and Samson, 1989) wag applied to deter-
mine which geographic areas systematically con-
tributed to above- and below-average fine particle
mass (PM2 5) over eastern North America (Brook
et al., 2004). Six-hour average measurements from
12 rural or suburban locations in eastern North
America, collected using the TEOM measurement
method, were individually associated with corre-
sponding 3-day back-trajectories for the warm sea-
sons (May through September) of 2000 and 2001.
Much of the populated area of northeastern
Canada and the United States was implicated in
the build-up of PM2 5 to "above average" concen-
trations (Figure 6.3). Average concentrations were
determined by calculating the mean concentration
at each of the sites during the warm seasons of
2000 and 2001. The finer structure of the QTBA
pattern indicated that transport from the Ohio
River Valley was most often associated with the
highest PM2 _ concentrations, particularly the
eastern portion of this area. In addition, air mass-
es traversing a relatively large area from southeast
Ohio to the western part of Virginia and the west-
ern Kentucky to central Tennessee area tend to
result in relatively high PM2 5 concentrations over
northeastern North America. These observation-
based findings are consistent with the spatial dis-
tribution of the major S02 and NOx point sources
(Figure 4.1a and Figure 4.3a).
6.1.1.4 Back-trajectory Analysis of PM2 5
Transport to Eastern Canada
Using hourly TEOM PM2 5 observations from
May-September Of the years 1998-2000, Brook et
al. (2002) have quantified the impact of various
transport directions on PM2 5 concentrations in
eastern Canada using back-trajectory analysis.
Comparisons of PM2 5 levels at different sites
reveal that on average, the local contribution to
total PM2 5 in the Greater Toronto Area is approx-
imately 30 to 35 percent. This implies that the
regional or long-range contribution comprises the
remaining 65 to 70 percent. Furthermore, at sites
in eastern Canada, average PM2 5 concentrations
were 2 to 4 times greater under south/southwester-
ly flow than under northerly flow conditions during
May through September of 1998 and 1999 (see
84

-------
chapter 6
Figure 6.3 'Combined OTBA
plot derived using 2OT0 and
2001 TEOM PM,j g measure^
ments for the warm months
(May-September). The locations
of the 10 measurement sites
(receptors) are shown by stars
and the locations of the maxi-
mum OTBA values are indicated
by the black circles. QTBA val-
ues greater than 1.0 indicate a
high likelihood of air masses
passing over that area bringing
above-average warm-season
PM2 s to the receptor.
Figure 2.2). This observation suggests that the
majority of PM2 s at these locations is arriving
from the transport of PM2 5 and PM2 5 precursors
from sources: south of this region.
6.1.1.5 Sources of PM to Glacier National Park,
Montana
Trajectory Clustering/Time Series Analysis was
applied to Glacier National Park in Montana
(Sirois and Vet, pers.Comm.J, This preliminary
analysis identifies the potential influence of west-
ern Canadian and U.S. sources to visibility impair-
ment at Glacier National Park. Qualitatively, S02
sources in Alberta, Saskatchewan, Montana and
North Dakota contribute to SOj-induced low visi-
bility events at Glacier National Park. High con-
centrations; of NO| observed at the Park were asso-
ciated with westerly air flow from the
Vancouver/Seattle area. Total OC and total BC, the
major contributors to visibility impairment at the
Park, were associated with air flows from the
VancOuver/Seattle, Oregon, and Northern
California areas.
6.1.1.6 Sources of PM and Acid Rain Precursors to
Southwestern Ontario: Study 1
Trajectory Clustering/Time Series Analysis was
applied to observed concentrations of particle
SO J and NO3 in air, and tojpH, SOj and NOg in
precipitation at the Longwoods measurement site
of the Canadian Air and Precipitation Monitoring
Network (CAPMoN) in southwestern Ontario to
determine source-receptor relationships (Vet and
Sirois, pers.comm,). The technique combined 3-
day back-trajectories with daily PM and ion meas-
urements, The technique involved categorizing the
air mass trajectories into two geographical sectors
(Figure 6.4) and sorting the data at the Longwoods
site according to the sector that each trajectory fell
within. The criteria for categorizing the trajectories
were as follows: 1) if at least 70 percent of the
points along the trajectory path fell within a sector,
the trajectory was categorized as originating from
this sector; and 2) if less than 70 percent of the
points along a trajectory fell within a sector, the
trajectory was categorized as "not attributable"
(N/A). Figures 6.5 and 6.6 illustrate the long-term
trends and median concentrations of particle SOj
85


-------
chapter 6
USA
N/A
1985
TNO," (ng/rrr)
Canada
U.S.A.
N/A
2848
7.39
8,14
1990	1995
SOj (ugW)
Canada	U.S.A.
S02 (lig/rrr )
T
2000
N 1945	2885	1527
G. Mean 1.74	4.96	3.59
Median 1.82	5.52	3.95
N 1923
G. Mean 2.62
Median 2,91
Canada
d)
N	1955
G Mean 178
Median	1.83
2000
Figure 6.5 - Long-term trends and median concentrations of SO^ (a and b, respectively), particle SOj (c and d,
respectively), and particle NOj (e and f. respectively) in air at Longwoods, Ontario associated with three-day back
trajectories from Canada, the United States and "Not Attributable* (N/A) to either sector. The trend line in the box
plots connects the geometric means, the line dividing the boxes .represents the median, the upper and lower sides
of the boxes represent the 75th and 25th percentile of the data, respectively and the upper and lower bars on
the box plots represent the 75th percentile plus 1.5 times the inter-quartile range and the: 25th percentile minus
1,5 times the inter-quartile range, respectively.
87

-------
Canada - United States Transboundary PM Science Assessment
PH
-ii|iiir-
a)!  v
'  ' - Sr c - V ' V A * r
*. ;j  " - f,'*'-;-y ;?'
U.S.A
b)
N 559
A. Mean 5.21
Median 5.10
	1
1985
1990
1995
2000
1556
4.38
4.25
I	
583
4.68
4.41

x
10 =
101
10
10"1
102
10"3
S04= (mg/l)
TIII	1I	1I	1	1I|	1-
rItftf]
USA
Nfit
i iIJ	i i rIi	lj	i I i	ii i I i
1985	1990	1995	2000
d).
N 625
G. Mean 1.65
Median 1.70
T
10
100 r
10"
10
1650
3.32
329
T
642
2.55
2.66
io=r	~	l
t i JLi
:?
t t
f



*
fr

Canada
USA.
N/A
N03" (mg/l)
10^
101
10
10"'
10"2
10"3
r
-iiii|iir
e)

k " 1 |L *( 	. r k ^ .r # 1* 9 -V"
: I - ** " "" " 
. Canada
	, USA	r
MA
I 1	I	I	l	E	I	I	I	I	I	I	I	I	I	l_
1985
1990
1995
2000
^ N	624
G Mean	1.58
Median	1.66
10
1645
3.34
3.22
639
3.43
3.40
r ;
r

' Xi
r	
r

r
pT Ti
+ ~ |
 1 i
U.S.A.
Figure 6.6 - Long-term trends and median concentrations of pH (a and b, respectively), SO^ (c and d, respectively)
and NOg (e and f, respectively) in precipitation at Longwoods, Ontario associated with 72-hour back trajectories
from Canada, the United States and "Not Attributable" (N/A) to either sector. The trend line in the box plots
connects the geometric means, the line dividing the boxes represents the median, the upper and lower sides of the
boxes represent the 75th and 25th percentile of the data, respectively and the upper and lower bars on the box plots
represent the 75th percentile plus 1.5 times the inter-quartile range and the 25th percentile minus 1.5 times the
inter-quartile range, respectively.
88


-------
Canada - United States Transboundary PM Science Assessment
Hg.m
20
10
9
8
7
6
5
4
3
2
0
-3
Hg.m
9.0
6.0
5.5
5.0
4.5
4.0
3.0
2.5
2.0
1.5
0.0
-3
TNOj"
(.ig.m
o.o
-3
Figure 6.7 - The geometric mean concentration of S02i SO4 and TNOJ measured in air at Lohgwoods, ON
(1983-2000) for the particular subset f air mass trajectories that passed through that grid square.
90


-------
Canada - United States Transboundary PM Science Assessment
ing to PM2 5 in Toronto and Vancouver respective-
ly (Figure 6,9). In Toronto, the main components of
PM2 5 identified were coal combustion (30 per-
cent) related to regional transport, secondary NOg
(34 percent) related to both local and upwind
sources of NOx and NH3, secondary organic
aerosols and biomass burning (9 percent) and
motor vehicle traffic (9 percent). Coal combustion
was related to regional transport (both from
Canada and the United States) as there are no sig-
nificant emission sources of coal combustion in
the immediate area. As a result, the signal detect-
ed using the PMF analysis is related to transport
into the area from non-local sources. The other
detectable components- were road salt (winter),
road dust/soil (yearly), smelters or related indus-
try, and oil combustion. In Vancouver, the three
major components were secondary NH4N03 (49
percent), secondary organic acid with SOj (23 per-
cent), and motor vehicles (20 percent). The minor
components were road dust/soil, sea salt and oil
combustion. The average PM2 5 mass in Vancouver
was observed to be approximately 44 percent
lower than PM2 5 levels in Toronto, The total influ-
ence of localized vehicle-related sources was esti-
mated to be 36 percent and 51 percent in Toronto
and Vancouver respectively.
6.1.2.2 Comparability of Receptor Model Results on
PM2 5 Sources in Toronto
The raw data from Lee et al. were subsequently
conveyed to a team of U.S. analysts at the Vermont
Department Of Environmental Conservation,
where they were analyzed using a second receptor
model, UNMIX. The independent PMF and UNMIX
results were then compared, refined and revised
with local surface meteorological data and ensem-
ble backward trajectory techniques then applied to
help evaluate and interpret the results. Annual
average PM2 5 mass contributions from the result-
ing PMF and UNMIX sources are displayed in
Figure 6.10.
The average annual PM2 5 mass concentration
during this period was 14 |ig/m3 (just below the
level of the U.S. standard) , with maximum 24-hour
concentrations (98th percentile) of 35 |ig/m3 (just
above the Canadian standard). Both models
reproduced the measured annual and daily PM2 5
mass measurements in terms of the identified con-
tributors, which included smelters (5 percent),
(a)
Toronto	H Summer 0Winter  Total
Ammonium Secondary Organic Vehicles	Road Road Salt Smelter	Oil
fJitrate	Coal	Acids	Dust/Soil	Combustion
)
Vancouver	~ Summer DWinter BTotal
Ammonium	Organic	Vehicles Road Dust/Soil	Sea Salt Oil Combustion
Nitrate	Acid s/Sul fate
Figure 6.9 - Percent contribution.
by component, to PMj j mass
observed in a) Toronto and
b) Vancouver as determined
using PMKMLR,
92

-------
chapter 6
Figure 6.10 - Annual average modelled PM ,
contributions in Toronto, (February 2000 - February
2001) using UNMIX (a) and PMF (b) receptor
modelling techniques
UNMIX Sources
[ 0 2nrtary MH4N03  Primary Coal
D Road Oust > MV BNaCI + WW
| ~ Road Oust	B Stricter i Asj
U 2nda*y (NH4J2S04 Snttary Acids
~ Gas MV CicsiHMV
 SirwHor (Zn>	
.an Pafc Mm Apt NMf Iim J J	Sip Del Mav 0:
 fctawVaf ckri  cruliis r: Co a	~ XfmttrurkNaiM
Figure 6.12 - Seasonal variations in Toronto PMF &
UNMIX source contributions.
250D 
i
 2000
1
Sun Won Tue Wed "ftiu Fri Sat Sun
 Diesel MV P Road Dust <,xSj n Gas MV |
Figure 6.11 - Average day of week variations in UNMIX
motor vehicle sources.
motor vehicles (20 percent), NH4N03 (36 percent)
and coal combustion (36 percent).
Figures 6.11 and 6.12 display the temporal
characteristics of several components, which pro-
vide insight into source characteristics. Figure
6.11 displays the average day-of-week contribu-
tions for the three identified UNMIX "motor vehi-
cle-related" sources. All three of these sources
decline substantially on weekends, with the rela-
tive reduction on Sundays being greatest from
diesel vehicles; least from gasoline vehicles, and
intermediate for road dust.
Figure 6.12 shows the seasonal patterns in
four major categories, shown for averaged PMF
and UNMIX results, The total influence from
motor vehicles and smelter sources is relatively
constant over the year, while the NH4N03 and
coal-related components Show strong winter and
summer peaks respectively
An evaluation of the receptor model daily
source contributions as a function of local surface
meteorology is illustrated for the UNMIX "motor
vehicle-related" and "coal-related" sources in
Figure 6.13. The influence of the mobile source
does not vary greatly with wind direction, but is
consistently higher for directions as wind speed
decreases (blue-shaded sectors), indicative of a
predominantly local origin. The coal-related
sources show a different pattern, all increasing
substantially with surface winds from the south
(southeast through southwest), and consistently
higher from this direction as wind speed increases
(red-shaded sectors], indicative of more distant
source influences.
Figure 6.14 (left panel) shows similar back tra-
jectory-based "incremental probability fields:"1 for
93

-------
Canada - United States Transboundary PM Science Assessment
Gasdine MV
MV
Pnrnary Cost
(NH4g8CM
Figured.13 - UNMIX motor
vehicle arid trial-related sources
vs. local surface wind speed
arid direction., Bluf shading
emphasizes directions from
which source influence-is
greatest at low wind speeds.
Pink shading emphasizes
directions where source
influence increases at all
wind speeds. Red shading
emphasizes directions from
which source influence,
increases at high wind speeds.
Upwind Probability Fields for 3 Upwind Probability Fields for "Coal Related"
"Coal Related" Toronto Sources Sources at 5 Northeastern Receptor sites
Underhili, VT,
ishington, DC,
Figure 6.14 - Incremental
probability fields for coal-related
sources at Toronto and other
eastern sites,
coal-related sources to Toronto PM2 5. This coal-
related source can be split into three separate
components based on the PMF and UNMIX analy-
ses: primary coal (emitted from the source in par-
ticle phase), secondary (NH4)2S04 and acidic Sul-
phates/secondary organics, While there is not a
perfect correspondence in their upwind probability
fields, the most probable upwind locations for all
three sources are similar and converge on a U.S.
region of high-density emissions from coal-fired
utilities. In the right hand panel of Figure 6.14, the
probability fields for the three Toronto "coal-related"
sources are combined and compared (at similar
incremental probability contours of 0.002) with
94

-------
chapter 6
Lye Brool
Waters
Great Smoky Mountains NP
Toronto NH4N03 Source
Increases 25% on
Weekdays vs. Weekends
Total Coal 2ndaryN03 Total MV
Averaged PMF and UNMIX Toronto Sources
Upwind Probability Fields for Ammonium Nitrate
Weekday Source Increases in Toronto
Figure 6.15 - Incremental probability fields and day-of-week patterns in Toronto NH4N03:"souks:"!
3 MV Source
Cwnponems
0,001
0.002
0.003
AmmoftLm
w urate
0,002
A/O.003
A aw
3 Coal Source
Compenerts
0.002
0.003
0.0M
Toronto Source Contributions by Daily PM-2.5 Mass Category
.
5 cd
frS S-10 10-20 20-30 30-H0 H0
PM-2.5 Mass Ccncenfratcns (ugAnS)
Upwind Probabrity Ftds for Major Toronto Source Categories
Figure 6.16 --Summary of major 1\yronto source regions and influences on daily PMa  mass concentrations.
similar results from other recent studies which
have applied a similar combination of PMF and or
UNMIX receptor models and ensemble back trajec-
tory techniques. The consistency and convergence
of results from these different model applications
adds confidence to the Toronto results, and sug-
gests a common "universal donor" source region
influencing multiple receptor locations in the
Northeast transboundary region.
Certain features of the modelled NH4N0-3
component also suggest a complex "causality".
The left side of Figure 6.15 compares the incre-
mental probability field for the Toronto NH4N0s
sources: with those from other recent receptor
modelling studies at (rural) eastern U.S. sites.
There is a moderately strong degree of conver-
gence in the most common upwind areas for high
NOj from these widely separated receptor sites,
which suggests a critical influence from agricultural
(fertilizer and livestock) NH3 emissions in the
north-central U.S. "corn belt." However, as indicated
on the right side of Figure 6,15, there is a moder-
95


-------
chapter 6
Figure 6.18 - Pie charts of the
source apportionment .results
for various locations in the
United States. (Some charts
moved for clarity)
E	KutefeCoiJ
aa	KibBb
Mod Is
^	Hotiue Buring
^	IndutWal
IE	CokM and SUI
~	Cmor/Not IdertHod
Back-trajectory analyses and wind/pollution
roses yield source location information for the
apportioned PM2 5 contributors. Nitrate sources
are associated with the Midwest farming regions
while the back-trajectory analyses for the oil-based
SOj component indicated large southern source
regions. The analysis for the SOj component is
complicated by the fact that some of the sources
seem to be related to high-pressure systems (as
evidenced by the clockwise swirl of many of the
back trajectories for the high source days).
Sulphate, from either eoal-or oil-based
sources, accounts for about one-third of PM2 5
mass. The next largest portion is either from NOj
components or mobile sources with all three of
these categories showing long-range transport
components. The smaller source:contributions are
more: site-specific, except for crustal dust. As-
many as eight source categories, including marine
influences, metal production, general industrial,
and fuel oil, are within the range of resolvability
with approximately one year of speciation data at
current levels of technology. Linking wind trajecto-
ries with the source apportionment results allows
one to develop source regions (i.e., geographic
regions with a high probability of being the origin
of the mass associated with a source profile).
These source regions provide evidence that at
least some of the particles associated with the
source profiles are likely transported over long dis-
tances. For example, the highest probability
source region for the coal combustion source pro-
file for Birmingham includes parts of the following
states: Missouri, Illinois, Indiana, Ohio, Kentucky,
Virginia, North Carolina, South Carolina, Alabama,
and Mississippi.
6.T.2.4 Compilation of PM2 5 Source
Apportionment Studies from the United States
The U.S. EPA summarized the findings of 27 source
apportionment studies covering over 30 locations.
The literature compilation found that contribu-
tions from secondary SOj and coal combustion
sources were the largest or one of the largest
sources of PM2 5 in nearly every study, often con-
tributing more than 50 percent of PM2 5 to the
receptor. Furthermore, these trajectory analyses
often pointed to source regions containing coal-
fired power plants. In addition, if the study time
frame was sufficiently long, secondary SOj and
coal combustion had different winter and summer
profiles- which were attributed to extremes of
97

-------
Canada - United States Transboundary PM Science Assessment
atmospheric chemistry between source regions and
receptors. Studies looking at longer time periods
observed reductions in contributions for some
sources (power plants, smelters), attributed to
reductions in emissions, fuel switching (from oil tO:
natural gas), and changes in meteorological condi-
tions (warm winters in late 1990s). For the western
locations, mobile sources and vegetative burning
tend to have larger contributions to total PM.
Figure 6.18 shows pie charts of the various appor-
tionment results for areas across the United States.
In general, the results from many of the stud-
ies were similar. A few receptors were studied
repeatedly, such as Underhill, Vermont, and
Brigantine, New Jersey, The contributors identified
are grouped into seven categories: SO=/coal,
mobile, NCfc biomass burning, industrial, crustal
and salt, and other/not identified. Note that in
Figure 6,18 the results from neighboring sites are
generally quite similar.
6.1.2.5 Source Locations and Time Series Analyses
in U.S. Cities
A number of studies have assessed sources of
observed PM2 5 in U.S. cities. In these studies,
PMF and UNMIX were either used individually or
in tandem to apportion sources to observed PM2 5
levels. All back trajectory analyses for sites in the
eastern United States associate the SOj compo-
nent of PM with the Ohio River Valley area. Several
studies noted transport across the Canadian bor-
der, specifically SOj from the midwestern United
States into Canada, and smelter emissions from
Canada into the northeastern United States. There
are plans to use the back-trajectory data to quanti-
fy the transport; however, these studies are not yet
complete. All of the studies looked at long-term
averages and most looked at seasonal (3-month)
averages. There was very little analysis of daily or
weekly events, with a few exceptions. (For the
most part, the studies considered are motivated by
long-term concerns, such as trends in regional
haze.) Lee et al. (2003 a) followed up on a crustal
source by identifying several days that were possi-
bly influenced by Saharan dust. Coutant et al.
(2002) mention the influence of fireworks in
Houston, Texas, Long (2002) studied a particular
event (2002 Winter Olympics) and documented
changes in the source proportions (mobile sources
were higher) and temporal changes (mobile
e 2002/0
r: 22
Figure:6,19 - The composites of MODIStder.ived aerosol optical depth (color) and cloud optical depth (black-white)
superimposed over continuous PM, g monitors (bars) for July 6th and 7th, 2GH2. The hourly PM2 5 mass concentration
is indicated by the height of the bar, while the color of the bar represents the 24-hour running average mass-
concentration color coded to the US EPA Air Quality Index The yellow to red colors of aerosol optical depth show
elevated aerosol concentrations and have been found to correlate strongly with PMa 6 levels. Note the elevated PM
associated with both measures of aerosol in Canada and the northeast United States,
98


-------










-------







-------
Appendix
will impact both ground-level ozone and particu-
late matter, and given that secondary PM2 5 pro-
duction and gas-phase photochemistry are closely
linked, it is natural to consider both pollutants
together.
A2.2 Model Evaluation for Winter 1998
Figure A.2 shows observed and modelled time
series of PM2 5 mass from February 7th until
February 14th, 1998 at three locations in south-
eastern Canada: Kitchener, Ontario (2a);
PM2.5
ug/m3
1
J July 1st 1995 Y /

PM2.5
ugm'i
1
July 1st 1995 | y ^
;w







A


\
?
-------
Canada - United States Transboundary PM Science Assessment
Kftchenef, Wes! Avenue Station (Ont.),
Modeled PM., 	
Observed PM,	
Hamilton Beasley Park Station (Ont.)
Modeled PMn 	 j|
Observed PMa	 'I
OS/02 IOJD2 11a 12/02 13/02 1-WJ2
00.00 00.00 00 00 00:00 00:00 00:00
072 08/02 09A32 10.-02 11/02 12/02
0000 00 00 0000 COOO 0000 00:00
Montreal. Drummond Station (Que l
Modeled PMrt 	
150
140
130
120
110
100
00
80
70
80
50
40
20
10
00 00
00:00
00 00
COOO
00 00
00:00
Figure A-2 - Time series of modelled and observed PM, s from February 7th to 14th 1998 for (a) Kitchener; Ontario-.,
(b) Hamilton; Ontario, and (e) Montreal; Quetee; Solid line: AURAMS simulation pgXntffi Dashed line;
Observations (jLig/m3),
Hamilton, Ontario (2b): and Montreal, Quebec
(2c), This period is of interest because of the
strong PM2 5 episode that occurred in eastern
Canada, beginning on February 9th in Kitchener
and Hamilton and on February 10th in Montreal.
Hourly PM2 5 concentrations as high as 70 |ag/m3,
were measured at the two Ontario stations while
the peak observed PM2 5 concentration for
Montreal was over 130 |ig/m3, AURAMS is able tO:
predict the observed day-to-day variation in parti-
cle mass over this one-week period, giving a good
representation of the mid-week increase in particle
mass., The modelled time traces are much
smoother than the continuous measurement
record, which can be attributed, at least in part, ta
the spatial resolution of AURAMS of 42 km. At this
resolution, AURAMS can only capture some Of the
features that are measured at a particular point,
due to fine-scale variations in meteorology and
emissions (i.e., point vs. grid-volume incommen-
surability). Note too that the TEOM measure-
ments are likely to be biased low, especially at
night, due to the impact of the heated inlet on
semi-volatile PM components entering the instru-
ment from wintertime ambient conditions.
Figure A.3 shows the corresponding ozone
time series plot for Hamilton. Ontario. The PM2 5
episode was reflected in the ozone concentration
time series by very low ozone levels, possibly as ,a
result of enhanced N02 titration associated with
1 18

-------
Appendix
Hamilton, Beasley Park Station (OnL)
Modeled Ozorw 	
Observed Ozor>e	
07i"02 0&02 09/02 10/02 11/02 12/02 13X12 14/02
0000 00:00 0000 0000 0000 0000 0000 00:00
Ctota
Figure A.3 s Time serif of modelled and observed
ozone (ppb) for Hamilton; Ontario from February 7th
to 14th 1998- Solid line: AURAMS simulation; dashed
linei observations.
stagnant conditions, AURAMS again tracked the
changes in ozone concentration very well.
Model-vs-measurement scatter plots of hourly
ozone and hourly PM2 5 concentrations are shown
in Figure A.4 for this same one-week period. On
the left, the ozone scatter plot shows a bias
towards under-prediction of hourly ozone concen-
trations (slope of 0.63,) and an R2 value of 0.48.
The PM2 5 scatter plot also shows an overall bias
to under-prediction (slope of 0.65), but some of
the smallest value can be grossly over-predicted,
driving the offset of the regression line to a value
of 10.5 |ag/m3 (see Table A.2). For PM2 5 the R2
value is 0,26. Note that hourly PM2 5 concentra-
tions as high as 160 |ag/m3 were both observed and
modelled for this period. A summary of the per-
formance statistics corresponding to Figure A.4 is
given in Table A.2.
Figures A.5 and A.6 present scatter plots com-
paring observed and predicted daily mass concen-
trations for PM2 5 and its three inorganic chemical
components for two sets of PM2 5 samples: the
IMPROVE and GAViM PM2 5 measurements were
reported for ambient conditions (Figure A.5)
whereas the NAPS PM2 s measurements were
reported at STP (Figure A.6). AURAMS PM predic-
tions are for .ambient conditions, but an STP con-
version was done in the preparation of Figure A.6
to allow a direct comparison with the NAPS data.
Note that particle NH+ was not measured by the
IMPROVE network for that time period but was
measured by the GAViM network. These plots are
characterized by relatively few measurements (see
Table A.3) and large scatter, particularly for the
three chemical components, but such scatter is:
characteristic of comparisons of air-quality model
predictions made for such short averaging periods..
Aside from one or tw outliers, agreement is bet-
ter than an order-of-magnitude, and there seems
to be a tendency to under-predict S04 and over-
predict NOs. Predictions of total PM2 5 mass do
not display much bias. Note too that the time aver-
aging associated with daily measurements rather
than hourly measurements does reduce scatter
(e.g., Figure A.4b vs, Figure A.6a). The performance
statistics corresponding to Figures A.5 and A.6 are
provided in Tables A.3 and A.4. Interestingly, a rel-
atively high R2 and a low RMSE are obtained for
the N03" PM component when comparing with
IMPROVE and GAViM measurements, whereas the
situation is reversed when comparing with the
NAPS observations. The number of speciated
NAPS data for that period is very limited, however,
which limits the meaningfulness of the statistics.
Table A.2 Performance statistics for hourly ozone and PM2 5 mass at all stations over the domain
from February 7th to 14th 1998. (where m number of observed values; R2: correlation coeffi-
cient; RMSE; root mean square error; NME: normalized mean error; MB: mean bias; NMB:
normalized mean bias (Kang et al., 2003))
Species n R2 aobs" RMSE NME	MB NMB	curve fitting equation
Hourly 03 34060 0.48 170 206 129 12.0 42.7	-5.2	-24.2	mod = 0.63 obs + 2.80
Hourly PM2 5 2739 0.26 582 360 232 22.0 61.4	2.8	13.1	mod = 0.65 obs +10.50
Legend: * variance of model data, "variance of observation data, ""covariance
1 19

-------
Canada - United States Transboundary PM Science Assessment
Model
Model
ieo	
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Hourly PM25 (ug/m3)	Obs
Hourly Ozone (ppb)	Obs
Figure A-4 - Scatter plots for hourly ozone: (left panel) and hourly PM,. ^ [right panel) at all stations over the domain for
all hours from February 7th to 14th 1998. Gray lines are 10:7, 1.1, 7:10 and 2:1, 1.1, 1:2 model-to-observation ratio lines
in left and right panels, respectively.
Table A.3 Performance statistics for PM2 5 mass:, PM2 5 S04, PM2 5 NH4, and PM2 5 N03
concentrations at all available IMPROVE and GAViM stations over the AURAMS domain
for all days from February 7th 00Z to 14th 00Z 1998. (see Table A.2 for definitions).
Species
n
R2
^mod
*obs
"mod-obs
RMSE
NME
MB
NMB curve fitting equation
PM2,s mass
31
0.23
297
19
36
15.9
67.1
3.2
29.0
mod = 1.87 obs - 6.37
PMj 5 SO4
30
0.02
1.8
3.4
0.3
2.4
59.4
-1.1
37.1
mod = 0.09 obs + 1.66
pm2.5 nh4
8
0.36
22
0.4
1.9
5.2
276.1
2.8
260.4
mod = 4.28 obs - 0.73
pm2.5no3
30
0.42
67
1.5
6.5
8.2
294.2
3.4
271.4
mod = 4.33 obs - 0.78
Legend: * variance of model data, "variance of observation data, ""covariance
120


-------
Canada - United States Transboundary PM Science Assessment
Particulate Matter fine fraction (ug/m3)
Sulfate fine fraction (ug/m3)
Ammonium fine fraction (ug/m3)
Nitrate fine fraction (ug/m3)
Figure A.6: - Same as Figure A5 but for Canadian NAPS-stations and Hg/m3 at STP.
Table A.4 Performance statistics for PM2 5 mass, PM2 5 SOj, PM2 5 NH j, and PM2 5 NOg concentrations
at all available Canadian NAPS stations over the AURAMS domain for all days from February
7th 00Z t-: 14th 00Z 1998. (see Table A.2 for definitions)
Species
n
R2
"mod
"obs
"mod-obs
RMSE
NME
MB
NMB curve fitting equation
PM3 5 mass
27
0.33
563
252
215
20.5
43.3
6.0
16.0
mod = 0.85 obs + 11.62
PM^ 5 SO4
7
0.24
2.3
4.1
-1.5
4.9
63.1
-3.9
-54.5
mod = -0.37 obs + 5.80
pm2.5 nh4
7
0.01
13
1.6
-0.5
5.2
78.4
3.6
67.9
mod = -0.29 obs + 9.76
pmz5 no3
7
0.01
126
15
-3.9
19.6
172.1
15.3
172.1
mod = -0.26 obs + 26.59
Legend: * variance of model data, ""variance of observation data, '"covariance
122

-------
Appendix
A2.3 Model Evaluation for Summer 1995
Figure A.7 presents time series erf modelled and
observed ground-level ozone for two stations in
the Windsor-Quebec City corridor for a ten-day
period July 8th to 17th, 1995. The left panel is for
a station in London, Ontario while the right panel
is for a station at St-Anicet, Quebec, west of
Montreal. This period includes both episodic and
London Aql Site (Ont.)
Modeled Ozone 	
Observed Ozone	
Sainl-Antcet (Ou6)
Modeled Ozone 	
Observed Ozone	
0
08/07
00:00
09/07
12:00
11/07
00:00
12/07
12:00
14/07
OOOO
15/07
12:00
17/07
00:00
09/07
12:00
11*>7
00:00
12/07
12:00
14/07
00:00
15/07
12:00
17/07
00 00
08/07
00:00
Date
Figure A-7 - Time series of modelled and observed ozone for London, Ontario (left panel) and Saint-Anieet, Quebec
(right panel) from July 8th to 16th 1995. Solid line: AURAMS prediction; dashed line: observations. All values in ppb.
Model
Model
Hourly PM25 (ug/m3)
Hourly Ozone (ppb)
Figure A-8 - Scatter plots for 6zsone at all stations over the domain for all hour's from July 8th to 11th 1995 (left panel)
and from July 12th to 15th (right panel). Gray lines are 10.7, 1:1 and 7:10 model-to-6bservation ratio lines. All values in
ppb.
123

-------
Canada - United States Transboundary PM Science Assessment
Table A.5 Performance statistics for hourly ozone at all stations over the domain from July 8th to 11th and
July 12th to 15th 1995. (see T&ble 2 for definitions)
Species	n R2 c^'	(Wous RMSE NME	MB	NMB	curve fitting equation
Hourly 03 (left) 40437 0.39 396 488 273 21.4 48.8	11.0	31.6	mod = 0.56 obs + 26.35
Hourly 03 (right) 39913 0.50 745 849 561 23.8 40.3	9.9	21.1	mod = 0.66 obs + 25.91
Legend: * variance of model data, "variance of observation data, ""covariance
non-episodic conditions: ozone concentrations
above 80 ppb were observed on July 13th and 14th
at both stations. Both panels show that the ozone
levels simulated by AURAMS during the worst days
of the period (July 12th to 15th) are in better agree-
ment with the observations than for the four pre-
ceding days. AURAMS over-predicts ozone at the
beginning of the period but improves with time as
the observed episode becomes more intense. This
behaviour can also be seen in Figure A.8, which
presents two scatter plots of modelled vs.
observed ozone. The left panel covers the pre-
episode period from July 7th to 11111 while the right
panel covers the period from July 12th to the 15th,
the four days with the highest ozone concentra-
tions. The two scatter plots are for all stations in
the domain at all available hours. The slope of the
best-fit line increases from 0.56 to 0.66 for the July
7-11 period vs. the July 12-15 period while the R2
value between AURAMS-predicted and observed
hourly ozone improves from 0.39 to 0.50. The com-
plete Set of performance Statistics is provided in
Table A.5. For the episode period, observed hourly
ozone levels reach over 180 ppb and predicted
levels reach about 150 ppb.
Figure A.9 shows an image of the AURAMS
predicted ground-level ozone field for July 14th
1995 at 2100 UTC at the height of the episode with
the observations for that time superimposed as
colored circles at the measurement locations. In
this figure, matching colours indicate good agree-
ment with the measurements. There is generally
good agreement between the model and the
observations. Observed and modelled values are
similar in the peak areas and we see the same pat-
terns in the modelled field as in the observations.
This figure is representative of the level of agree-
TTzone"
Figure A.9 - AURAMS simulated ground-level o^one for
2100 UTC (1700 EOT) on July 14th 1995. Observations
valid.for this hour are repr^spnted by the circle-
Values are in ppb.
ment between model and observations for the
afternoon and early evening period. Figures A.7
and A.9 lead us to conclude that AURAMS repro-
duces the afternoon and evening portion of the
diurnal cycle of ground-level ozone well but tends
to over-predict during the night and morning
hours.
Figures A. 10 and A. 11 show scatter plots com-
paring observed and predicted daily mass concen-
trations for PM2 5 and its three inorganic chemical
components for the eight-day period from July 8th
to the 15tfl, As noted above, the number of meas-
urements available for PM2 5 comparisons is con-
siderably less than that available for ozone,
despite the fact that additional measurements
124


-------
Canada - United States Transboundary PM Science Assessment
Ammonium fine fraction (ug/m3)
Nitrate fine fraction (ug/m3)
Mode
Particulate Matter fine fraction (ug/m3)
Model
10
Model
50
45
40
35
30
25
20
15
10
5
0
0
Model
16
5 10 15 20 25 30 35 40 45 50
Sulfate fine fraction (ug/m3)	obs
Figure A. 11 - Same as Figure A. 10 but for Canadian NAPS-stations and pg/ffl3 at STP.
Table A.6 Performance statistics for PM2 5 mass, PM2 5 SOj, PM2 5 NH j, and PM2 5 NOj concentrations
at all available IMPROVE, GAViM and CAAMP stations over the AURAMS domain for all days
from July 8th 00Z to 16th 00Z 1995. (see Table A.2 for definitions)
Species
n
R2
^mod
obs
^mod-obs
RMSE NME
MB
NMB
curve fitting equation
PM2 5 mass
66
0.26
70
94
42
9.1
38.5
-1.0
-5.8
mod - 0.44 obs + 8.74
PM2 5 SO4
32
0.60
65
10
20
9.1
110.4
6.8
108.4
mod = 1,97 obs + 0.66
pm25 nh4
1
-
-
-
-
-
-
-
-
-
PM2 5 NO3
32
0.01
0.12
0.10
-0.01
0.5
105
-0.2
-65.6
mod = -0.08 obs + 0 14
Legend: * variance of model data, "variance of observation data, "'covariance
126


-------
Canada - United States Transboundary PM Science Assessment
comparisons with observations, AURAMS pro-
duces reasonable simulations of particulate SOj
and NHJ, but the level of agreement for particulate
NO3 is less satisfactory. As explained in the previ-
ous section, however, N03~ measurements also:
have a much higher degree of uncertainty, especial-
ly for periods as far back as 1995. However,
AURAMS's ability to simulate the concentrations of
the various inorganic PM components is also simi-
lar to what has been reported for other aerosol and
oxidant models (e.g. Mebust et al., 2003), including
the lower performance skill for particulate N03~.
Finally it should be kept in mind that the evalua-
tion of AURAMS was: focused on two relatively
short periods, therefore little to no averaging was
done when comparing with observations.
Figure A. 12 - Comparison of AURAMS KSS^J output
with satellite imagery. Top: Visible satellite image
valid at 2SOS UTC on July 10th 1995, Bottom: AURAMS
PM2:5 output in the lower levels valid at 2SQ0- UTC.
128


-------

-------


-------